@Schoasch
Basic information
Quantitative crypto backtesting & Bitcoin-cycle analytics. DSR-corrected, point-in-time, look-ahead-aware — proprietary, not from public APIs. Free tier.
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"backtesting-arena": {
"url": "https://tradingstrategies.work/api/mcp",
"headers": {
"Authorization": "Bearer sk-arena-..."
}
}
}
}How hot is the Bitcoin market today? Daily 0-100 heat score for the Bitcoin market, aggregated from 8 components (BTC-Cycle, F&G, Altcoin-Season, Bullmarket-Ampel, Funding-Rate, Hash-Ribbons, Mayer-Multiple, MVRV-Z). Returns score, band label, color, 7d/30d delta, verdict, components breakdown, plus score_percentile ranking today’s score against its own history (e.g. 42 = 44th percentile — how hot/cold vs history, not just the raw number). [Free tier]
Crypto cycle position — where are we in the cycle? Default BTC: point-in-time 10-indicator aggregation (MVRV-Z, NUPL, Puell, Pi-Cycle, Funding, Hash-Ribbons, Power-Law, Rainbow, F&G, Mayer). Pass asset=ETH or asset=SOL for a per-coin cycle read built from the transferable price-derived indicators (Mayer, weekly-RSI, 200-week-MA distance) with renormalized weights; BTC-native indicators (halving, dominance, mining, hash-ribbons, F&G, Pi-Cycle, on-chain) are explicitly returned as `not_applicable` rather than faked. All return raw + Z-Score, signal enum, and a `percentiles` block ranking each indicator against that asset’s own history. The `signal` enum is a FIXED SCORE-BAND LABEL (<25 accumulation · 25–45 recovery · 45–60 expansion · 60–75 distribution · ≥75 overheated), not an independent market-phase detection — a mid-band score reads "expansion" even in a drawdown market (the 45–60 band is the neutral middle; cross-check price/drawdown context before quoting the label as a market state). BTC additionally returns `highlights[]` (rule-based markers for currently unusual indicator values — descriptive, versioned ruleset; empty array = nothing unusual) and `price_context` (price at scoring time vs live spot with drift % — the scores are based on the scoring-time price, not the live spot). Point-in-time scored — not reconstructable from a generic price API. Note for volatility questions: this tool carries the regime context around a volatility reading (Funding, Mayer, Pi-Cycle) but not the volatility series itself — that is arena_get_volatility_history. Related: arena_get_historical_analog (what followed states like this one), arena_get_bullmarket_ampel, arena_get_pulse. [Free tier]
Current BTC, ETH and SOL spot price — what is Bitcoin (or ETH/SOL) worth right now? Live USDT-quoted last price plus 24h change %, high and low from Binance. Use this to anchor the connector’s own analytics (cycle, historical-analog, gem scores) with the current market price instead of switching to web search mid-analysis. [Free tier]
Aggregate stablecoin supply (crypto-liquidity proxy) — is the liquidity impulse turning or accelerating? macro_regime only gives the 30d delta; this exposes the trend: current supply, 30d/90d change (USD + %) plus a compact time series so direction and speed are visible, not just a single delta. Read `impulse` for what the supply change is doing — four states (accelerating / decelerating / reversal / flat). The neighbouring `acceleration_usd` is the signed difference last-30d minus prior-30d and gets LARGE exactly when the trend reverses, while the older boolean `accelerating` requires the same direction AND a bigger magnitude; a reversal therefore shows a big `acceleration_usd` next to `accelerating: false`. Source DefiLlama peggedUSD. [Free tier]
Spot-ETF net flows (USD millions) — is the flow impulse turning or accelerating? The summary only gives point-in-time deltas; this exposes the trend: 30d/90d net flow, a direction label (inflows/outflows/flat) and a compact cumulative-inflow time series so direction and speed are visible, not just a single delta. Read `impulse` for what the flow is doing — it has four states (accelerating / decelerating / reversal / flat) and is the field to quote. Two neighbouring fields measure different things and are easy to confuse: `acceleration_usd_m` is the signed difference last-30d minus prior-30d and gets LARGE precisely when the flow reverses, while the older boolean `accelerating` requires the same direction AND a bigger magnitude — so a swing from outflows to inflows shows a big positive `acceleration_usd_m` together with `accelerating: false`, which is correct and reads like a contradiction. `impulse` reports that case as 'reversal'. Default BTC; pass asset=ETH or asset=SOL. Source SoSoValue. [Free tier]
Is it altcoin season? Daily Altcoin-Season indicator (v7 Native-Filter methodology). Returns BTC-Dominance, Alt-Dominance, 4 Layer-1 signals (USDT.D, USDC.D, BTC-DOM, ETH-DOM), overall color (red/amber/green) + Top-50 CoinGecko snapshot. [Free tier]
How fearful or greedy is the market right now? Crypto Fear & Greed Index (alternative.me). Returns the current `value` (0-100) and `classification` (extreme fear / fear / neutral / greed / extreme greed) as their own fields, plus `history` — the last 90 daily readings by default, so you can see whether today is a move or a plateau. The index is a contrarian-read sentiment gauge, not a timing signal: it says where sentiment stands, not what price does next. The window is capped and the `range` block states requested / granted / available days with the reason — a short series here is a window, not a young index. On Pro and Elite the `cadence` block adds our own sentiment-momentum measure (how far smoothed sentiment has travelled versus ~90 days ago) — a figure the upstream index does not publish; on Free the block is present but its value is null with a stated reason. For the regime around a reading use arena_get_cycle; for what followed comparable sentiment states use arena_get_historical_analog(preset="deep_fear"). [Free tier · cadence Pro+]
Is this still a bull market? Bitcoin Bullmarket-Ampel current state (0-5 active stages). Returns active_count, a stages[] breakdown (each stage with key, label, active and `since` = first day of its current state; null when the state predates the 400-day lookup) and stage_history (active_count of the last 30 days). Higher count = more bull-market signals firing. Stages evaluate weekly 20W/50W-MA conditions. [Free tier]
Are longs or shorts paying right now? Latest BTC perpetual funding rate, averaged across up to three exchanges (Binance, Bybit, OKX; 8h settlement cadence). Returns value, 30d moving average and Z-Score. Positive = longs pay shorts (bullish bias), negative = shorts pay longs (bearish bias). Read `coverage` before comparing values across dates: it says how many exchanges stand behind that day (3 = full average, 1 = a single exchange), and a day-over-day move can be a change in composition rather than in the market. [Free tier]
Are miners capitulating? Latest Hash Ribbons indicator (Charles Edwards). Returns 30d and 60d hashrate moving averages — when 30d > 60d after a capitulation, signals miner recovery (bullish). [Free tier]
Is BTC stretched against its 200-day average? Returns the current Mayer Multiple — BTC price divided by its 200-day SMA — as spelled-out fields: date, mayer_multiple, price_usd, sma_200d (with sma_200d_source declaring its provenance) and the Trace Mayer (2014) bands as a machine-readable field (<0.7 capitulation, 0.7–1.5 neutral, 1.5–2.4 bullish, >2.4 euphoria). The former raw row shape (d, value.v, value.close) is still present but deprecated — see deprecated_fields with removal date. One ratio, not a regime call: for the multi-indicator cycle read use arena_get_cycle, for the series behind this number arena_get_mayer_multiple_history. The bands are historical description, not thresholds to trade. [Free tier]
What is the macro backdrop doing? Daily Macro Regime snapshot from 18 components in 6 tiers (Liquidity 30%, Financial Conditions 20%, Risk Appetite 15%, Crypto Liquidity 10%, Business Cycle 15%, Inflation/Real Rates 10%). FRED-sourced. Returns composite_score (0-100), regime_label (risk_off/neutral/risk_on_leaning/risk_on), cycle_phase_label (contraction/early_expansion/mid_expansion/late_expansion), matrix_quadrant (sweet_spot/late_cycle_warning/crisis/recovery), tier_scores (6 sub-scores), components (flat key/value of all 18), plus stale_components_detail dating each stale input (last_good_date + age_days + discontinued flag for series the upstream has retired for good) so freshness is quantified, not a vague caveat. Two component keys mean something narrower than their name suggests, so read them carefully: `vix_score` is the derived 0-100 score (a value of 71 means VIX around 18.6), NOT the VIX index level — the raw Cboe level is not redistributed over this channel; and `broad_dollar_index` is FRED DTWEXBGS (Broad USD Index, Jan 2006 = 100), NOT the ICE DXY, so readings near 120 are normal. The former names `vix` and `dxy` are still present with identical values but are deprecated and listed in deprecated_fields with their removal date. [Free tier]
What does Bitcoin actually move with? Pre-aggregated weekly correlations between Bitcoin and 13 macro components (Fed Net Liquidity, VIX, DXY, Real Yield 10Y, NFCI, Yield Curve, etc.). Returns quadrant_performance (BTC return stats per 2D-matrix quadrant — annualized return, vol, max drawdown, positive-period%), component_correlations (Pearson 90d/1y/5y per macro component + quartile-performance), asset_correlations (Pearson per window + per quadrant; assets: dxy plus tokenized on-venue proxies paxg = PAX Gold, spyb = S&P 500 ETF proxy, qqqb = Nasdaq-100 ETF proxy — proxies carry tracking noise vs. the underlying, and windows the vehicle history does not cover are null with data_start_date telling you why: the ETF proxies listed on Binance mid-2026, so their windows fill in over time — 90d first, ~2 months after listing), current_quadrant. Window labels are upper bounds — sample_size_days / data_start_date carry the actual basis. Historical analysis over the windows named above. [Free tier]
Is the trend up or down, and how fresh is the flip? Daily Bitcoin market structure from 1000-bar Phantomflow adaptation (BTCUSDT 1d). Returns current_trend (up/down/sideways), last trend change timestamp, counts of waves + fractals, last-5 fractals on each side (up = pivot highs, down = pivot lows), and trend_context: previous trend + its duration, flip_age_days, and a descriptive historical flip base rate over the SAME 1000 bars (total flips, share reverted within 5 bars, median trend duration) — a fresh same-day flip is the least reliable observation, the base rate says how often such flips reverted historically; it is NOT a forecast for the current flip. Educational analysis of price action. [Free tier]
Which price levels matter above and below spot? Reproducible Bitcoin structural levels on BOTH sides of spot, in TWO distinct provenance classes. (1) resistance/support: swing-pivot clusters — where past pivot highs+lows cluster into price zones (touch-count, band, last-touch date, signed distance), resistance above spot, support below, nearest-first. (2) indicator_levels.above / .below: named indicator STANDS as marks — 200-day & 200-week simple moving averages, short-term-holder cost basis, Pi-Cycle legs — each carrying its source, formula and as_of date. The two classes are kept separate on purpose: pivots are where price REACTED before, indicator levels are where an indicator STANDS now. Both are measured price clusters: they say where trading has concentrated, not where anyone defends a level. [Free tier]
What is the options market pricing in? Latest Deribit volatility snapshot for BTC or ETH. Returns DVOL (30d vol index), constant-maturity ATM implied vol (30/60/90/180d via options chain), 30d realized vol, and `vol_risk_premium_30d`, which is the TRAILING spread: ATM implied vol (30d, from the options chain — not DVOL) minus the realised volatility of the PAST 30 days. It answers "are options priced expensively right now?". Set include_implied=true to additionally get the FORWARD premium in an `implied` block: DVOL(t) minus the realised volatility of the FOLLOWING 30 days, which answers the different question "did the expectation actually materialise?". These two are NOT interchangeable — measured across 1,037 paired days they carry OPPOSITE signs on 17.3% of days for BTC and 30.5% for ETH. The forward field is spelled out as `vol_risk_premium_forward_30d` so the two cannot be confused. The most recent 30 days carry premium_complete=false and no premium value at all, because their forward window has not closed yet; they are excluded from every aggregate. Descriptive context, not a trading signal. Source: Deribit DVOL Index. History: BTC from 2021-04-01, ETH from 2022-02-15. [Free tier]
How did market heat get to where it is? Returns the Arena-Pulse TIME SERIES: one row per day with date, 0–100 score and band, in ascending date order. Use it for trend, turning points and "how did we get here"; for today's value alone call arena_get_pulse (cheaper, one row). Range capped by tier. [Free 30d / Pro 365d / Power unlimited]
How has the cycle score moved over time? Returns the BTC-Cycle TIME SERIES: one row per day with adj_score and z_adj_score, ascending by date. The scores are point-in-time — each day carries the value computed from data available on that day, so the series can be used for look-ahead-free analysis. For the current cycle reading alone call arena_get_cycle; for what similar historical readings were followed by, call arena_get_historical_analog. Range capped by tier. [Free 30d / Pro 365d / Power unlimited]
Has capital been rotating into or out of altcoins? Returns the Altcoin-Season TIME SERIES: one row per day with overall_color, BTC / alt / stablecoin dominance and BTC price, ascending by date. Use it to see whether capital has been rotating into or out of alts over time; for today's state alone call arena_get_altcoin_season. Range capped by tier. [Free 30d / Pro 365d / Power unlimited]
How has leverage positioning shifted over time? Returns the BTC perpetual funding-rate TIME SERIES, aggregated across the available perpetual venues (Binance, Bybit, OKX), at the 8h funding cadence, ascending by date. The response is SEGMENTED by venue composition: each segment covers a stretch with a STABLE venue set (venues, coverage, from/to, its rows); composition_breaks lists the transitions with a mechanically derived cause (venue_added / venue_removed / venue_changed — e.g. OKX joined 2026-01-28). Averages, z-scores or percentiles computed ACROSS segments mix different venue populations — compute within a segment, or accept the mix knowingly; composition_stable: true means the whole window is one segment and safe to treat as one series. Positive funding means longs pay shorts (crowded long positioning) and vice versa; sustained extremes are a positioning signal, single prints are noise. For the latest funding value alone call arena_get_funding_rate. Range capped by tier. [Free 30d / Pro 365d / Power unlimited]
How stretched has BTC been against its 200-day average? Returns the Mayer-Multiple TIME SERIES (BTC price ÷ its 200-day SMA), one row per day, ascending by date. Values around 1 mean price sits at its 200d average; historically high readings clustered near cycle tops and low ones near bottoms — a descriptive ratio, not a trigger. For the current value alone call arena_get_mayer_multiple. Range capped by tier. [Free 30d / Pro 365d / Power unlimited]
Which on-chain series are available? Lists all 40 available Bitcoin Research Kit (BRK) on-chain series across the groups pilot, sentiment, mining, cohort, cointime, activity (e.g. MVRV, NUPL, SOPR, Realized-Price, Mayer, Puell, STH/LTH SOPR, Hash-Ribbons). Returns id + label + group. Use the id with arena_get_onchain_latest / _history. [Free tier]
What does this on-chain metric read right now? Returns the most recent value of ONE on-chain series from the Bitcoin Research Kit as { series_id, metric_name, date, value }. Cheapest way to answer "what is X right now" (MVRV, SOPR, realized price, hash rate, …). Discover valid series_ids with arena_list_onchain_series; for the history behind the number use arena_get_onchain_history. A single reading has no context — pair it with the series percentile before calling any level high or low. [Free tier]
What did recent buyers pay on average? Latest BTC short-term-holder cost basis (realized price of coins younger than ~155 days, BRK brk_sth_realized_price) plus derived STH-MVRV (spot ÷ STH cost basis) and an in_loss flag (spot below cost basis = recent buyers underwater in aggregate, historically stress / near local bottoms). Descriptive on-chain context, not a buy/sell signal. [Free tier]
How has this on-chain metric moved over time? Returns the full TIME SERIES of one on-chain metric from the Bitcoin Research Kit — date/value pairs in ascending order, with history back to 2009 for most series. Use it for trend and percentile work; for the single current reading call arena_get_onchain_latest, and to discover valid series_ids call arena_list_onchain_series. Values are as-reported: on-chain metrics can be revised retroactively, so this is not a point-in-time vintage. Range capped by tier — the response carries a `range` block (requested_days, granted_days, clamped, clamp_reason, tier), so a clamped window announces itself instead of silently looking like the full history. [Free 30d / Pro 365d / Power unlimited]
What happened at the last Deribit expiry? Max pain and how spot settled against it: max_pain_strike, spot_at_expiry, %-diff, put_call_ratio, notional. Plus up to 10 upcoming expiries, each with current live max-pain level, days_to_expiry and open_interest_contracts. On days_to_expiry, mind the clamp: it is floored at 0 and therefore CANNOT tell "expiry is today, still hours away" from "expiry was today, already settled" — the same reading covers a live state and a post-mortem. Use settles_at (full ISO timestamp of the settlement moment) and hours_to_settlement (SIGNED — negative means already settled but not yet finalized, so the row is still listed here) whenever that distinction matters. settlement_time_utc names the settlement time assumed for this market; it is 08:00:00Z for DERIBIT_BTC, measured against the exchange itself (public/get_instruments: 868 of 868 open instruments across 13 expiry dates, daily through quarterly, all exactly 08:00:00 UTC). For a market whose settlement time is not evidenced, all three fields are null rather than guessed — a precise-looking wrong timestamp would be worse than none. Read the OI: a max-pain level is only as meaningful as the open interest behind it — a daily expiry with 2,000 contracts and a quarterly with 154,000 are not the same observation. `oi_available` distinguishes "null" from "not collected". Upcoming expiries also carry open_notional_usd together with notional_spot and notional_spot_date. The two contract fields are the SAME measurement at different observation times, not two different quantities: open_interest_contracts on an upcoming expiry is the open interest at the LATEST daily snapshot, total_contracts on a settled one is the open interest at the LAST snapshot BEFORE expiry (verified in the cron: total_contracts is filled from lastSnap.total_oi_contracts). contracts_as_of names the snapshot those contracts came from. Where it equals expiry_date, contracts and settlement price are same-day — measured on all 64 finalized expiries, so a vintage mix inside total_notional_usd is structurally possible but has never occurred; the field exists so that a future cron gap becomes visible instead of passing silently. Mind the deliberate name split: total_notional_usd on a settled expiry is computed against the SETTLEMENT spot and never changes again, while open_notional_usd uses the CURRENT spot and therefore moves with spot every day, even if not a single contract trades. Same formula, different quantity — which is why the reference spot travels with it. Every expiry — upcoming ones included — now carries is_monthly, is_quarterly and expiry_type (daily | weekly | monthly | quarterly). These NEST rather than partition: quarterly ⊂ monthly ⊂ weekly ⊂ daily, so a quarterly expiry also has is_monthly=true. Filter on the booleans, read expiry_type as the label — it is the only field that separates a Friday expiry from a mid-week one (measured: Fri 18,613 contracts vs. Sat 2,852, and both used to read as false/false). All three are derived from the calendar (last Friday of the month; quarterly in Mar/Jun/Sep/Dec), which is why upcoming expiries can carry them at all — they have no settlement row yet. Cron collects daily 02:00 UTC from Deribit Public API. Related: arena_get_max_pain_history (base rates + daily snapshots of open expiries), arena_get_iv_snapshot (implied vol for the same expiries). [Free tier]
Does max pain actually pull price to the strike? Settled Deribit BTC options expiries with the max-pain level we compute per expiry, for measuring the convergence question: does spot drift toward the max-pain level as expiry approaches? Each row: expiry_date, max_pain_strike, spot_at_expiry, %-diff, P/C ratio, notional, expiry-type flags. With include_open_snapshots=true it adds the daily observation series of still-open expiries — that series starts 2026-05-28, is not backfillable, and its per-expiry depth is thin, so check open_snapshot_coverage before computing anything from it. Days auto-capped by tier: Pro 365d, Power 3650d. Max-pain levels are our own aggregation across the option chain; the chain itself is not redistributed. Source: Deribit. Related: arena_get_max_pain (current + upcoming), arena_get_iv_snapshot. [API Pro tier]
Do the two data sources still agree? Daily drift log comparing bgeometrics (legacy) vs BRK (canonical-soon) pilot metric pairs. Returns mean / max / outlier counts per pair for the requested window. Used by BRK-migration review (every 4 weeks). [API Pro tier]
Which strategies can I backtest here? Lists all backtest strategies (key, label, plan, supported asset classes, primary indicators). Filterable by asset class and plan. Use this before calling arena_run_backtest to discover valid strategy names. [Free tier]
Which asset universes can I test against? Lists all crypto asset universes (BTC, top-10 crypto, top-50 crypto, etc.) — the underlying pair-sets used by custom-report and universe-backtest endpoints. [Free tier]
Which pairs are in this universe? Returns one pair universe in full: its id, label, selection rule and the complete list of pairs it currently contains. Use it to see what you are about to test BEFORE handing a universe_id to arena_run_universe_backtest, or to resolve a universe into explicit pairs. For the list of available universes call arena_list_universes. Universes reflect the CURRENT membership — they are not point-in-time, so a backtest over them carries survivorship bias for the earlier years. [Free tier]
Which strategy and interval combinations actually performed? Aggregated backtest performance per (strategy × interval) cell. If `strategy` AND `interval` provided, returns detail with per-asset breakdown + param variants. Otherwise returns the matrix. Free tier is limited to the same strategies that are free in the backtester itself (rsi_sma, golden_cross, rsi_ob_os, bnh_fixed, dca_reference); the response then carries `plan_capped: true` plus `plan_cap_note`, so a short matrix is never mistaken for a thin database. Detail mode on a Pro-only strategy returns 403 rather than a silently empty answer. API Pro and Power receive every cell. [Free: 5 strategies / Pro+: full]
Do entry filters help, and which ones? Lift analysis of entry filters (200WMA, Altcoin-Season, ATR-Volatility, Bullmarket-Stage) per strategy combo — baseline vs filtered CAGR/win-rate/drawdown. [API Pro tier]
Does this strategy work better in calm or wild markets? Breaks realized strategy performance down by VOLATILITY PHASE (low / normal / high) per asset and timeframe, so you can see whether an edge only exists in one volatility regime. Answers "when does this work", not "does this work" — for the overall verdict use arena_get_strategy_insights, for the macro-regime cut arena_get_strategy_performance_by_regime, and for the raw volatility time series arena_get_volatility_history. Cells below min_trades are suppressed rather than shown as noise. [API Pro tier]
What are people backtesting right now? What is being backtested on Backtesting Arena right now — platform activity, NOT market sentiment. hotAssets, hotStrategies, trendingUp/trendingDown, assetDistribution, strategyAssetMatrix and totalRuns aggregate over the window selected by `period` (7d default, 30d or 90d). Three fields have FIXED windows independent of `period`: dailyActivity (daily counts, up to 365 days — year heatmap), weeklyTrend (weekly counts, up to 365 days), profitTrend (share of profitable runs per pair, last 30 days vs. previous 30 days). Honesty note: the counts include our own bulk and admin snapshot runs, so this is coverage-weighted attention, NOT a clean crowd signal — never present it as 'traders are bullish on X'. For actual market sentiment use arena_get_fear_greed, arena_get_funding_rate, arena_get_altcoin_season or arena_get_pulse. Replaces arena_get_sentiment (deprecated alias, removed after 2026-11-15). [Free tier]
DEPRECATED — renamed to arena_get_platform_activity: this endpoint measures PLATFORM ACTIVITY (what is being backtested on Backtesting Arena), not market sentiment, and the old name made consuming LLMs misuse it. Same payload plus deprecation fields; this alias will be removed after 2026-11-15. Use arena_get_platform_activity. For actual market sentiment use arena_get_fear_greed, arena_get_funding_rate, arena_get_altcoin_season or arena_get_pulse. [Free tier]
How did this exact strategy, asset and interval perform? Aggregated backtest performance for ONE specific (strategy, asset, interval) combination. Returns run_count, avg_cagr, avg_win_rate, avg_drawdown, effective_years, vs_buy_hold comparison (beats_buy_hold, cagr_delta) and an `evidence` block declaring the gate machine-readably (gate_applies_to: stats.run_count, threshold 5 runs, benchmark value, aggregation data window). For multi-strategy overview use arena_get_strategy_insights. Use this to answer 'How does strategy X perform on asset Y?'. [Free tier]
What would each entry filter have changed for this strategy? Per-(strategy, asset, interval) filter-effect analysis. Returns baseline-stats (no filters) + each observed filter-variant's stats with cagr_delta / drawdown_delta / win_rate_delta vs the time-overlap-matched baseline + best_by_cagr pick + not_applicable_filters list (e.g. altcoin_season excluded on BTC-pair). Baseline and each variant carry their aggregation `window` (from/to + avg_run_years) — CAGR is time-normalized, so identical trade sets over different windows legitimately produce different CAGR. Based on REAL backtest aggregations — not theoretical 2^5 permutations. Use this to answer 'Which filters would improve my backtest for X on Y?'. [Free tier]
In which macro regime has this strategy worked? Historical backtest performance for ONE (strategy, asset, interval) combination SPLIT BY macro market regime (sweet_spot / late_cycle_warning / crisis / recovery — classified at each trade's entry date), PLUS the CURRENT live regime so you can align the buckets yourself. Answers 'WHEN has this strategy worked?' — deliberately NOT 'should I trade now': the former recommendation.verdict was removed (2026-08-15) because it ranked regimes on the pooled trade sum and could flip when another user's backtest changed the pool. Each regime bucket returns trades, trades_per_config (trade counts pool ALL parameter-variant configs — see config_count), win_rate, avg_pnl_pct (per-trade return, not annualized), reward_risk_ratio (per-trade mean/stddev, NOT annualized Sharpe), share_of_time_pct (calendar-day-weighted — each regime observation counts the days until the next one, so the mixed weekly/daily cadence of the regime history does not skew the share) and a rating. The `benchmark` block anchors the payload with the combination's buy-and-hold CAGR (identical to arena_get_strategy_performance vs_buy_hold — without that anchor, regime avg_pnl_pct is a trajectory, not an excess). For a decision-grade view compose with arena_get_strategy_filter_effect and arena_is_distinguishable. [Free tier]
Which entry filter carries a real edge? Platform-wide aggregated analysis: how each Pro+ entry filter (200 WMA, ATR low/high/expansion, Altcoin Season, Bullmarket confirm/strict) affects strategy CAGR — baseline vs. filtered, asset-equal-weighted (per-asset medians over param-deduplicated runs, then the median across assets — no single asset's run grid can dominate an arm). delta_cagr is the median of PER-ASSET deltas over MATCHED assets only (present in both arms) — so it usually differs from filtered_cagr − baseline_cagr; pairs_matched/pairs_filtered and the baseline pairs count declare the basis. Verdicts come from the effect's 90% paired-bootstrap interval (delta_ci_low/delta_ci_high), not the point estimate: helps (whole interval > +1pp) / hurts (< −1pp) / neutral (inside ±1pp) / insufficient_evidence (runs disagree) / insufficient_data (fewer than 30 runs per arm or fewer than 10 matched assets). Below the gate, derived fields (delta_*, dsr, dsr_pass) are null; every gated null carries its reason (dsr_pass_reason, *_net_reason); the envelope `evidence` block declares the gate's referent and threshold machine-readably. Response is GROUPED by strategy: envelope fields (market, computed_at, n_trials) once, per strategy one baseline block {cagr, net_cagr, sharpe} plus filter cells; filter cells with zero runs are folded into filters_without_data. A full market is a few hundred cells — use limit/offset (strategies per page) plus the truncated flag for partial reads. Filters evaluated in isolation (no stacking); net values are median CAGR after per-side trading costs (verdict/delta stay gross). [Free tier]
What are the strongest backtest results on the platform? Public leaderboard: the highest-CAGR backtest results across all users, with anonymized usernames, pair, strategy, interval and period. Answers "what has scored best on this platform so far". Read it as a selected extreme, not as a recommendation — a top-of-leaderboard entry is the winner of a large search and its edge is upward-biased; arena_get_robustness_field or validate_strategy tell you whether a given result holds up. For the user's own runs use arena_list_backtests. [Free tier]
Which backtests have I run? Lists the backtest runs belonging to the authenticated user — newest first, with id, strategy, pair, interval, date range and headline metrics per run. Use it to find a run_id, then call arena_get_backtest for its detail or arena_get_backtest_trades for the individual trades. Only your OWN runs; for the public cross-user leaderboard use arena_get_winners. Paginated via limit + offset. [API Pro tier]
What exactly did that backtest do? Returns the full record of ONE backtest run by id: strategy, pair, interval, date range, parameters, filters and the aggregate metrics (CAGR, total return, win-rate, max drawdown, trade count, Buy & Hold comparison, net-of-fees figures). Only your own runs (admins may read others). Get ids from arena_list_backtests; for the individual trades add arena_get_backtest_trades; to create a new run use arena_run_backtest. [API Pro tier]
Which trades did that backtest actually take? Returns the individual round-trips of one of your backtest runs: entry and exit date, entry and exit price, per-trade P&L and the running equity after each trade. Use it when the aggregate metrics are not enough — to see whether a result rests on a handful of outlier trades, how long positions were held, or where the drawdown came from. Needs a run_id from arena_list_backtests; for the aggregates alone use arena_get_backtest. Closed round-trips only — a position still open at the end of the period is not counted. [API Pro tier]
What is in this shared backtest link? Fetches a backtest that someone published via a share link, addressed by its share_id — no ownership and no Pro tier required, which is what makes it the right tool when a user pastes a /shared/backtests URL. Returns the same result shape as arena_get_backtest (config plus aggregate metrics). Use arena_get_backtest instead for the user's own runs. [Free tier]
Is this strategy signalling buy or sell right now? Current signal-status (green/yellow/red) for a strategy on a pair+interval. Backed by the daily check-signals cron — needs at least one user with an active Ampel on this combination. [Free tier]
Does this strategy survive an honest test? Backtest a trading strategy honestly — look-ahead-aware validation with Deflated-Sharpe-Ratio / multiple-testing correction (Bailey & López de Prado). Returns an EVIDENCE verdict (insufficient_evidence | anecdote | failed_oos | passed_oos) plus metrics, flags and caveats — NOT a buy/sell recommendation. Call this before acting on a strategy or signal list. Accepts a named catalog strategy (type=rules), a timestamped BUY/SELL signal list (signal_list), or a timestamped trade list (trade_list). Checks: realistic next-bar fills (look-ahead/optimism), net of cost, out-of-sample split, and a hard 30-round-trip sample gate (under 30 is always "anecdote"). Not reproducible via generic backtest tools that ignore overfitting. [API Pro tier]
How would this strategy have performed? Run ONE strategy on ONE pair over a date range and get the full result: CAGR, total return, max drawdown, win-rate, trade count, Buy & Hold comparison, net-of-fees figures, and a run_id for later retrieval. Synchronous, typically 3–10s. Use this when the user wants a concrete result for a specific setup. For several strategies side by side use arena_compare_strategies; for many pairs at once use arena_run_universe_backtest; to judge whether an EXISTING result is trustworthy rather than produce a new one, use validate_strategy or arena_get_robustness_field. Filters are optional and only remove entries; run once without them for the baseline. Read result.benchmark before comparing cagr to buyhold_cagr: warmup or a late listing can shorten the strategy window, and matches_strategy_window:false means the two figures are annualized over DIFFERENT periods — in that case benchmark.strategy_window carries the like-for-like buy-and-hold over the window the strategy actually traded, and THAT is the one to compare against. A backtest is evidence about the past, never a forecast or a recommendation. Per-day quota: Pro=50, Power=500. [API Pro tier]
Which of these strategies performed best on the same data? Run 2–5 strategies against the SAME pair, interval and date range and return per-strategy metrics plus a comparison summary (best by CAGR, best by win-rate, worst by drawdown). Use this when the user asks which of several strategies fits a market — it holds the pair, interval and requested date range fixed, which a series of separate arena_run_backtest calls does not guarantee. What it does NOT equalize is the EVALUATION window: a strategy with a long warmup starts trading later, so compare actual_date_from across the runs and check result.benchmark before ranking by CAGR. For one strategy across many pairs use arena_run_universe_backtest instead. Caveat worth passing on: comparing N strategies and reporting the winner IS multiple testing — the winner’s edge is upward-biased. arena_get_robustness_field puts a counted N on that. Sequential, expect 10–50s. Per-day quota: Pro=20, Power=200. [API Pro tier]
Does this strategy hold up across a whole universe? Runs it against every pair in the universe. Pair cap depends on your API tier: Pro 50, Power 250 — Power therefore covers crypto-top-250 in ONE job instead of five, which matters because five jobs mean five separate result sets you have to merge by hand, and merging across different pair sets is exactly how a ranking ends up measuring pair selection instead of strategy quality. THIS CALL IS ASYNCHRONOUS AND RETURNS NOTHING BUT A job_id: the result is NOT in this response. You MUST poll arena_get_job_status until status is 'completed'. Budget the wait — background runtime is roughly 1.5 s per pair, so 50 pairs ≈ 1–2 min and a 250-pair job ≈ 6 min; estimated_seconds in the create-response gives the current estimate. Provide either universe_id (call arena_list_universes) OR explicit pairs[]. Benchmarks bnh_fixed and dca_reference are accepted here — run one of them over the SAME universe and interval before reading any result, because an excess over buy-and-hold is not a statement without the buy-and-hold value itself: measured on 41 common pairs, bnh_fixed sits at +0.2 % while the naive figure suggested −22 %, and a strategy 'beating' a −54 % benchmark means 'do not own this asset', not 'this strategy is good'. NOTE ON THE BUY-AND-HOLD COUNT: beat_buyhold_count compares each pair's cagr against its buyhold_cagr, and the two are not always measured over the same window — a strategy with a long warmup (or a pair listed after date_from) starts trading later, while the benchmark runs from the requested start. Treat the count as a tally, not a verdict, and check actual_date_from per pair. NOTE ON PERSISTENCE: universe results live ONLY in the job response (api_jobs.result). They are deliberately not written to backtest_runs, so they carry no filter_binding and no coin-denominated history, and you will not find them later via arena_list_backtests — copy what you need out of the job result. Per-day quota: Pro=5, Power=50. [API Pro tier]
Is my universe backtest finished? Polls an async job by job_id (created via arena_run_universe_backtest). Returns status (pending/running/completed/failed), progress_pct, pairs_completed, and once completed: the full result (summary + per-pair results). [Free tier]
Would a grid bot have made money here? Simulate a GRID BOT (buy-low / sell-high ladder inside a fixed price range) on historical candles. Returns final value, return %, CAGR, trade count, fees paid and a Buy & Hold comparison. This is a different machine from the strategy backtester: grid bots earn from oscillation inside a range, not from trend — for signal-based strategies use arena_run_backtest instead. The result depends heavily on the range you choose (low_price / high_price); a range the price left early makes the bot idle, so treat range choice as part of the hypothesis, not a detail — arena_suggest_grid_range proposes a defensible range. Each run is saved to your account (the returned id is the run_id); publish a public snapshot page with arena_share_grid_backtest. Free tier limited to BTCUSDT/ETHUSDT. Per-day quota: Free=5, Pro=50, Power=500. [Free / Pro / Power tier]
Which price range should my grid bot use? Answers the question arena_run_grid_backtest deliberately leaves open (its own description says: treat range choice as part of the hypothesis). Returns TWO independent range suggestions side by side: iv_anchored (BTCUSDT/ETHUSDT only — sized from option-implied volatility, Deribit DVOL Index; the methodically grounded answer) and recent_volatility (7 trading days of realized volatility before anchor_date — the method exchange auto-modes use, available for all pairs). When both exist and disagree strongly, the options market expects a different volatility regime than the recent past showed — that disagreement is information. Output field names (low_price, high_price, grid_count, grid_type) map 1:1 onto arena_run_grid_backtest inputs. Inputs: pair (required), anchor_date, sigma_mult (IV range width in sigmas, default 1.5), dte (IV horizon in days, default 30). For a historical backtest set anchor_date to your start_date so the volatility range matches the entry, not today; iv_anchored always reflects the latest IV snapshot regardless of anchor_date. Free tier limited to BTCUSDT/ETHUSDT (same gate as the grid backtest itself). [Free tier]
Want a public link for a grid result? Publishes a read-only snapshot page (with OG preview image) for one of YOUR saved grid backtest runs and returns its URL. Pass the run_id you got back from arena_run_grid_backtest (every authenticated run is saved automatically). The page shows the numbers the engine actually computed — this tool takes no result values, so shared pages cannot carry fabricated figures. Shares are permanent snapshots; language picks the page copy (en default). [Free tier]
What would a custom report cost? Get a pricing quote for a custom report (universe-backtest PDF + Excel) without committing to a purchase. Returns price, universe size + preview, excluded pairs, and filter config. Crypto universes use top-N tiers (top-10 … top-250) or a custom pair list. [API Pro tier]
Is my report ready? Poll the status of a Custom-Report job. Lifecycle: pending_payment → queued → running → generating → success/failed. Returns progress_pct, succeeded/failed counts, plus pdf_url / xlsx_url when done. [API Pro tier]
Notify me when this signal flips? Fires when an existing Ampel-Config's signal flips (BUY ↔ SELL). Prerequisite: the user must have created that ampel-config in the web UI (`/dashboard/ampel`) — pass its UUID here; this tool cannot create one. Optional signal_types filter narrows to BUY-only or SELL-only. For the current signal state without subscribing, call arena_get_signal_status. Creates a standing subscription; it does not return a value now — collect fired updates with arena_check_subscription_updates (polling) or receive them by webhook, and end it with arena_cancel_subscription. For the CURRENT value instead of a change notification, call the matching read tool. [API Pro tier and up — max 3 active subscriptions for Pro, 20 for Power]
Notify me when the cycle band changes? Bands: (capitulation → risk-off → neutral → constructive → euphoric). Optional bands filter restricts to specific target bands. For the current band without subscribing, call arena_get_cycle. Creates a standing subscription; it does not return a value now — collect fired updates with arena_check_subscription_updates (polling) or receive them by webhook, and end it with arena_cancel_subscription. For the CURRENT value instead of a change notification, call the matching read tool. [API Pro tier and up — max 3 active subscriptions for Pro, 20 for Power]
Notify me when market heat crosses a threshold? Fires when the daily 0–100 Arena-Pulse score crosses threshold_above (upward) or threshold_below (downward). At least one threshold is required. For the current score without subscribing, call arena_get_pulse. Creates a standing subscription; it does not return a value now — collect fired updates with arena_check_subscription_updates (polling) or receive them by webhook, and end it with arena_cancel_subscription. For the CURRENT value instead of a change notification, call the matching read tool. [API Pro tier and up — max 3 active subscriptions for Pro, 20 for Power]
Notify me when the bull-market stage count changes? Tracks the Bullmarket-Ampel active stage count (0–5). Optional direction filter (up/down/any) plus specific stages of interest. For the current stage count without subscribing, call arena_get_bullmarket_ampel. Creates a standing subscription; it does not return a value now — collect fired updates with arena_check_subscription_updates (polling) or receive them by webhook, and end it with arena_cancel_subscription. For the CURRENT value instead of a change notification, call the matching read tool. [API Pro tier and up — max 3 active subscriptions for Pro, 20 for Power]
Which alerts do I have running? Returns every ACTIVE subscription belonging to the current API key: id, type, trigger configuration, delivery method and expiry. Use it to see what is already running before creating a duplicate, and to get the subscription_id that arena_cancel_subscription needs. Does not return fired updates — that is arena_check_subscription_updates. [API Pro tier]
Has anything I subscribed to fired? Returns all undelivered updates for the API key, then marks them as delivered. Call regularly to consume the polling queue. Updates contain payload with subscription_type, current value, previous value, and trigger context. [API Pro tier]
Stop this alert? Deactivates one subscription by id, so it stops firing and frees a slot against the per-tier limit. Returns the deactivated subscription. Idempotent — cancelling an already-cancelled one is a no-op, not an error. Get ids from arena_list_subscriptions. Undelivered updates already queued are not removed. [API Pro tier]
Altcoin screener ranking — which altcoins look strong right now? Today's CoinGecko Top-200 scored by a composite of 3 factor groups: Mean-Reversion (A), Tokenomics (B), Market-Structure (C). Backtest-validated factors, not a hype list. Limit gated by tier: Free top-10, Pro top-50, Power top-200. [Free tier, daily refresh]
How does this altcoin score? Returns the Altcoin-Screener score for ONE coin, addressed by its CoinGecko id: the composite score, its group breakdown and — for Pro+ — the 9 raw factor values across groups A/B/C. Use it once a candidate is known; to rank or filter the whole screened universe use arena_get_gem_scores (plural), and for how the score behaved out-of-sample use arena_get_gem_validation. The score ranks relative attributes, it is not a price forecast or a buy signal. [Free tier]
Did the screener picks actually beat BTC? Bi-weekly equal-weight basket backtest for the screener picks vs BTC and market average. Shows CAGR, max drawdown, win-rate. NOTE: curves are precomputed weekly on an N ladder (10/25/50/100/200, same input snapshot and engine per rung); a requested N snaps to the nearest rung — `top_n` in the response names the rung actually used, `top_n_requested` echoes the request, `basis_note` declares any snap. [Free tier]
How volatile has Bitcoin been? Daily Bitcoin volatility time series: realized volatility (30d & 90d, √252-annualized, close-to-close) and ATR% (Wilder EMA-14, captures intraday range + gaps), on the same scale. Ranks come in two flavours and they answer different questions — `rvRank`/`atrPctAnnRank` expand from the start of history and are look-ahead-free, but BTC volatility has fallen structurally, so a filter like "rank below 10" mostly picks up that decline rather than a regime; `rvRankRolling`/`atrPctAnnRankRolling` rank against a trailing 2-year window and are the ones to use for cross-epoch regime comparisons. History reaches back to 2009 via a stitched pre-Binance close series; ATR is null before the Binance era because no daily high/low exists that far back (see meta.coverage). Use `from`/`to` for a specific window instead of pulling everything and discarding it, and `granularity`/`fields` to keep long ranges affordable. Agents fetching long ranges should pass `schema_version: "2026-08"` today: it rounds floats. Two savings figures, and they are not the same number. ON ITS OWN it cut 30–36 % of characters depending on window length (measured 2026-07-31; the saving falls on long ranges because the pre-Binance years carry null ATR, and nulls do not round). COMBINED with `fields: "minimal"` and `meta: "minimal"` it cut about 45 % (measured 2026-08-17). Budgeting from the single-effect figure therefore understates what the combination buys. Both are dated measurements, not promises — every response carries a `size` block with `chars_before`/`chars_after`/`saved_pct` for YOUR call, so read that instead of these numbers. It is opt-in until the default flips 2026-11-01. Free tier: last 365 days. Related: arena_get_volatility_phases (current phase per pair), arena_get_iv_snapshot (implied vs. this realized — same RV method, but its realized_vol_30d is computed at snapshot time BEFORE that date has traded, so on fresh breakout days the two can differ; this series uses completed closes and is the one to trust for finished days), arena_get_cycle (regime context). [Free tier]
Is this pair calm or wild right now? Current ATR-based volatility phase (low/normal/high/expansion) per tracked pair, updated daily at 08:00 UTC. This is a single current state — for the time series behind it use arena_get_volatility_history, and for what the phase implies for strategy choice use arena_get_volatility_recommendations. Filter with `pair` when you only care about one asset instead of pulling all of them. [Free tier]
Which strategies suit the current volatility phase? Top-3 by historical win-rate for that phase on a given pair. Phase comes from the latest snapshot (arena_get_volatility_phases); minimum 20 trades per phase required for inclusion. Answers "which strategies did well in a phase like the current one?" — a historical ranking of what held up in comparable phases. Related: arena_get_volatility_phases (the phase itself), arena_get_edge_reports (filter effects with verdicts), validate_strategy (evidence check on a concrete configuration). [API Pro tier]
What knowledge objects exist here? Discover what Knowledge Objects exist: lists all published types + their subjects (with min_tier, api_path, seo_slug, latest as_of). Use this BEFORE arena_get_knowledge to learn valid type/subject pairs instead of guessing. New types appear automatically. [Free tier]
What does the platform know about this subject? Fetch a versioned, explainable Knowledge Object by type + subject (e.g. type='market_regime', subject='GLOBAL'). Returns the current published envelope: payload, explanation (factors + weights + confidence), provenance (inputs + params), ontology binding, compute version. ONE tool covers ALL knowledge types. Set include_graph=true to also walk the knowledge graph: resolved outbound edges (what this object is derived_from / references) + inbound edges (what derives from / references it), each with api_path + seo_slug so you can follow them. [Free tier; per-object access additionally gated by min_tier]
What does this term mean here, exactly? Resolve a knowledge-platform term to its canonical definition (e.g. term='regime'). Returns label, definition (EN/DE), calculation, unit, source + source_ref, version, related terms. Use this to resolve the onto:<term>@<version> references inside Knowledge Objects. [Free tier]
Buy now or wait for the dip? Decision-math over the user's OWN assumptions (target/dip prices, probabilities, capital). Two modes: "compare" = expected value of Buy-Now vs Wait vs Split + the breakeven dip probability (prices as MULTIPLES of today); "allocate" = the risk-adjusted (Kelly / risk-aversion γ) optimal fraction to deploy now vs reserve for the dip (ABSOLUTE prices). Ask the user for the missing inputs, then call. Returns scenario numbers and which option wins on expected value — NOT a buy/sell recommendation. For the full interactive version (incl. leverage & Elliott-wave planning) point the user to https://tradingstrategies.work/analyse/dip-decision. [Free tier]
What happened historically after the Bitcoin cycle looked like this? Conditional forward-return distribution for a named preset cycle state — over N DISTINCT historical episodes matching that state (matched_episodes), returns median/IQR/positive-share forward returns (30/90/180/365d) with per-horizon n, small-n warnings, point-in-time integrity and an `evidence` block that names which field its sample-size gate checked (gate_applies_to), against which threshold, over which data window. A distribution with its sample size. Not obtainable from web search or public market-data APIs — requires point-in-time indicator history and look-ahead-free episode matching. Presets: cycle_bottom_cluster (Cycle bottom cluster), cycle_top_cluster (Cycle top cluster), deep_fear (Deep fear), euphoria (Euphoria), quiet_volatility (Quiet volatility regime). The response opens with "preset_definition" — the machine-readable condition set behind the preset name, plus current_state_matches (does the state hold TODAY?) and last_matching_date; check that before mapping the distribution onto the present. Some presets carry a "study_finding" field — a state we have already investigated and where the result was NULL; read that before the distribution, it is the more important answer. EVERY preset returns "vs_unconditional_drift" — read THAT rather than the raw forward returns: the raw median measures the asset's contemporaneous drift as much as the state — the drift and excess columns in that block carry the honest comparison, and the excess can be negative while the raw median looks positive. Also works for asset=ETH/SOL (F2 cycle history), but only price-derived presets (cycle_bottom_cluster, cycle_top_cluster) — fear-greed and volatility presets are BTC-only. Related: arena_get_volatility_history (the series behind the volatility preset), arena_get_cycle (the current state to compare against), arena_dip_scenario (composes this base rate into a tranche structure). [API Pro tier]
Where would I add on a dip, and when is the thesis wrong? Frame a dip/accumulation thesis WITHOUT a recommendation. Given an asset (BTC/ETH/SOL), a named cycle-state preset and a thesis horizon, returns: (1) a tranche LADDER anchored to STRUCTURAL marks (200-week MA, support clusters) below spot — not calendar-DCA, not a price forecast; (2) the cited historical base rate from the analog engine (what forward returns followed comparable states, with episodes_matched and small-n warnings); (3) the explicit lump-sum-vs-tranche tradeoff (laddering buys lower timing variance, NOT higher expected value). Requires an invalidation point (mandatory: at what scenario is the thesis wrong). Composes the historical-analog + key-levels tools. This structural framing is MCP-only; a related (different-method, EV/Kelly) interactive tool is at https://tradingstrategies.work/analyse/dip-decision. [API Pro tier]
Is this backtest result real, or a lucky cell? Assess one backtest result against its neighborhood instead of trusting a single "+X% CAGR" cell. Given a (strategy, interval, pair) and YOUR result (user_cagr, optional user_sharpe), returns: the cross-asset distribution of the SAME strategy+interval across every pair the backtest factory ran it on (median, IQR, positive-share, your percentile), a plateau/spike/fragile/mixed verdict, and — where Sharpe coverage allows — a Deflated Sharpe threshold whose N is COUNTED (the number of neighbor assets IS the testing family), not guessed. Honest small-n handling: fewer than 15 neighbors → "insufficient", no DSR-N claimed. Set axis="parameter" for the secondary, always-anecdotal view (the few parameter settings tested on this exact pair). Read-only over result aggregates, look-ahead free. [API Pro tier]
Do these two CAGR figures actually differ? Check before ranking them. Pass the two values as `a` and `b` (gross CAGR in percent, same basis) plus `axes` — which arbitrary choices went into them — and the tool returns whether their gap clears the MEASURED noise floor of those choices, along with the floor itself, the dominant axis, and the probe + date it was measured on. `axes` accepts: grid_phase (how a multi-day candle grid is aligned to the Unix epoch; exists only on 2d/3d), parameter_choice (neighbouring parameter settings — by far the largest at ~17.9 pp), window_edges (shifting the start date), pair_selection (which pairs made it into the universe). Pass ALL axes that genuinely varied; the floor is their maximum, not their sum. Optionally set `interval` to the candle interval so the floor can be sharpened where an axis was measured per interval — passing grid_phase together with a non-multi-day `interval` is a hard error, because that axis does not exist there. `label_a` and `label_b` are optional display names for the two values and are echoed back inside the explanation, so a multi-way comparison stays readable. Worked example: 2d vs 3d intervals differ by 4.40 pp, but grid alignment alone spans 6.66 pp — so that comparison carries no finding at all. Read-only, no market data touched. [Free tier]
Where does price sit inside its trend channel? Mechanically fitted trend channels for a pair: log-linear regression over close plus 1σ/2σ bands, computed over three windows at once (90/180/365 bars). The three windows are not selectable by design — a single window invites trying them until one supports the thesis, three side by side show whether a channel is robust or an artifact of the window choice, and `agreement.consistent` states which it is. The differentiator is `r_squared_percentile`: "R² 0.42" says nothing, "R² 0.42 — 31st percentile of all same-size windows on this pair" says this channel is worse defined than two thirds of past ones, which is what stops a wish-line being read as structure. Where history is too short the window is omitted and listed in `windows_unavailable` with a reason — never estimated from fewer bars; below 500 bars (1d) / 150 (1w) the raw values still come but `percentile` is null rather than a rounded number from too small a sample. Pairs listed within the last few months (the tokenized equities and ETFs) therefore return `data_sufficient: false` and an empty channel list — that is the answer, not an error. Set `interval` to '1w' for the weekly view; note that a window is counted in bars, so 365 on '1w' means 365 weeks and most pairs do not reach it. What a band edge is: a description of past dispersion, not a level the market defends. Related: arena_get_key_levels (pivot clusters), arena_get_btc_market_structure (trend flips and their base rate), arena_get_historical_analog (whether a condition like the current one ever paid). [Free tier]
What do the classic indicators read right now? Current RSI(14), MACD(12/26/9), Bollinger(20,2), ATR(14) and OBV for a pair — each with a PERCENTILE RANK against that indicator's own history on that pair, plus the observation count. The rank is the point of the tool: "RSI 43.2" is a number any charting package gives away, "RSI 43.2, 24th percentile of 2,808 observations" is a placement. ATR comes as a percentage of price so it is comparable across time, and OBV as a 30-bar slope normalised by that window's volume, because raw cumulative OBV would mostly rank how long the series has existed. Where the reading sits in an extreme AND a study on this platform has tested that exact state, the payload carries the study verdict — including a null result: a Bollinger squeeze returns the `quiet_volatility` finding that tight bands did NOT carry an edge across 47 episodes. Below 500 bars (1d) / 150 (1w) the raw values still come but `percentile` is null with a reason, rather than a rounded number from too small a sample. Set `interval` to '1w' for the weekly view. On the 1d view the payload also carries `rsi_14_weekly` (weekly RSI with its own rank) — for BTCUSDT this is the SAME series as arena_get_cycle rsi_weekly, measured character-identical (its source_note carries the measurement). What the labels mean: `state` (oversold/neutral/overbought) names where a reading sits on its own scale — a description of a level, not an instruction. Related: arena_get_trend_channels (structure), arena_get_historical_analog (did a condition like this one ever pay?), arena_get_volatility_history (the volatility series behind ATR). [Free tier]
How far above or below its moving averages did price stand back then? A measured time series, not an estimate. Covers six MAs (50/100/200-day and 50/100/200-week; weekly MAs are simple averages over ISO-week closes, the same definition arena_get_cycle uses, so today's value agrees with that tool). Per date and MA: distance_pct plus TWO percentile ranks — expanding against the pair's own full prior history (with rank_n, null below 500 observations) and rolling over the trailing 504 days. Warm-up discipline: before an MA's full window exists the field is null, never an average over fewer days — a "200-week MA" computed from 400 days looks plausible and is a different metric; the coverage block names each MA's first valid date. BTCUSDT reaches back to 2011-07 via the platform's own pre-Binance closes (source named in price_source); the first valid 200-week value is ~2015-05. Range capped by tier from today backwards; granularity daily/weekly/monthly, coarsened with an explicit note when a request would exceed the row cap. What the numbers are: ratios and ranks, not absolute price or MA levels. For today's absolute MA levels as named marks use arena_get_key_levels (indicator_levels block); for the cycle-scored 200-week distance use arena_get_cycle. [Free 30d / Pro 365d / Power 3650d]
Should I take this entry? Answers it for one (strategy, pair, interval) in ONE call instead of seven. Aligns what each entry filter historically did to this strategy (arena_get_strategy_filter_effect) with where that filter stands TODAY (bull-market gauge, altcoin-season signal, volatility phase, 200-week trend for BTC): `filters[].blocks_this_entry` says which filter would sit this entry out, with the measured worst-loss / return deltas next to it. Adds the current signal state (anticipated is always false — before candle close there is no signal), an `edge_vs_benchmark` block gated by the MEASURED noise floor (a gap below the floor is a measurement artifact, not a finding), a `contradictions` block (e.g. Pulse risk-off while the macro regime reads risk-on — reported, never resolved), and measured invalidation zones (pivot clusters, 200-week SMA; BTC only). `detail`: 'headline' (default) returns the statement, three key numbers and only the decisive filters; 'full' adds every variant, the raw pulse/macro/filter-effect blocks. Every source can fail independently — sources_used / sources_unavailable make the basis auditable; the answer never silently narrows. Returns a plain-language `statement` with its `confidence` and the reason for that confidence — state it, do not hedge it further; the payload carries its own scope note. Compose further with arena_get_strategy_performance_by_regime (WHEN has this worked) and arena_is_distinguishable. [Free tier]
Check for additional tools whenever your task might benefit from specialized capabilities - even if existing tools could work as a fallback.
Backtesting Arena is an MCP server for quantitative crypto backtesting and market-regime analytics. It delivers proprietary outputs—such as DSR-corrected strategy evaluations, look‑ahead‑aware backtest validation, and point‑in‑time Bitcoin cycle scores—that are not reproducible from public OHLCV or market‑data APIs. The platform provides evidence, distributions and base rates, never buy/sell recommendations.
Authenticate using a Bearer token or claude.ai one‑click OAuth, then interact via the three available channels: REST, MCP, or x402. All 68 tools are visible during discovery; calls to Pro/Power tier tools from the free tier return a structured tool‑error with an upgrade_url.
validate_strategyIts core outputs—DSR‑corrected strategy evaluations, look‑ahead‑aware validation, point‑in‑time 10‑indicator Bitcoin cycle scoring, macro‑regime composites, and Edge‑Library filter effects—are proprietary and not reproducible from public OHLCV or market‑data APIs.
No. The platform delivers evidence, distributions and base rates only; it never provides buy/sell recommendations.
Yes, rate limits are enforced at the API‑key level.
All tools are visible during discovery. When a free‑tier call hits a Pro/Power tool, it returns a structured tool‑error with an `upgrade_url`—there are no hidden tools, and escalation is transparent.
Either a Bearer token or claude.ai one‑click OAuth.
Serve your docs to AI agents over SSH and MCP. Agents use ls, cat, grep, find, and pipes against a virtual filesystem backed by your documentation. Identity-scoped access, governed writing with human approval, and portab
Local tools that let your AI use Mail, Calendar, Contacts, Teams, Slack, WhatsApp, OneDrive, Google Drive, Notion, Outlook and Office files on Mac and Windows. Runs on your computer.
Medplum is a healthcare platform that helps you quickly develop high-quality compliant applications.
Open-source Node.js FHIR MCP server with SMART Backend Services, metadata-aware search/CRUD tools, compact responses, FHIRPath filtering, safe pagination, audit events, and terminology lookup.
Catch intent conflicts before code conflicts. The open-source coordination protocol for coding agents, built above Git.
Comments