fundamental_arb_exec dossier — does the short-horizon Black-Scholes edge survive executable entry prices, OOS, net?

Status: FALSIFIED under the pre-registered rule (results 2026-09-14, commit 7542179). Executable fills remove the fundamental_arb_short edge; not promoted. Pre-registered 2026-09-14 as a separate hypothesis, after the fundamental_arb_short post-hoc diagnostics pointed to a stale-price artifact and before any fundamental_arb_exec result was computed.

Pre-registration

Frozen with the G4c checks in tests/g4c/. Nothing below may change after a result exists; a change means a new, separately registered hypothesis.

Hypothesis. For Polymarket markets that ask whether an asset's price will be above, below, or touch a strike by a date, a Black-Scholes fair probability (archived fundamental_arb pricer, realized volatility) identifies mispriced YES prices. Taking the model's side when the gap is large earns a positive net return out of sample and beats a no-skill benchmark under the same costs.

This is a limited test. Volatility is realized volatility for every asset. The archived bot preferred ATM implied volatility for equities; that variant is not tested here.

Universe.

• Markets parsed by the archived parser archive/oddsoddy/strategies/fundamental_arb/fundamental_arb_strategy/scanner.py::parse_market, imported verbatim.
• endDate between 2026-04-10T00:00:00Z and 2026-08-21T09:07:27Z, closed true, and a clean resolution: oddsx.data.markets.resolve_outcome gives YES = 0 or 1.
• Sides above, below, touch_above, touch_below. between is excluded (the archived config disabled it).
• Symbols: BTC/USD, ETH/USD, SOL/USD, AAPL, MSFT, GOOGL, AMZN, NVDA, TSLA, META, SPY.
• Strike sanity: 0.5 ≤ strike / spot at decision ≤ 2.0. This removes index-level strikes mapped to SPY and non-price "index" markets.

Decision time. One decision per market.

• t_dec = max(first_trade + 10 minutes, end − 1 hour), rounded up to the whole second.
• Skip the market if end − t_dec < 30 minutes.
• Skip the market if no trade priced it in the 10 minutes before t_dec.

YES price at decision. Price of trades in [t_dec − 10 min, t_dec), weighted by token_amount.

• YES is the first token in clobTokenIds.
• A trade whose outcome token is the second token is converted to YES price as 1 − price.

Entry (fill) price. The signal still uses the YES price above; the bet is filled at an executable price.

• Fill trades: the market's trades with the earliest timestamp in [t_dec, t_dec + 10 min).
• fill_price: those trades' YES prices (second-token trades converted as 1 − price), weighted by token_amount; fill_ts = that timestamp.
• No fill trade in the window, or a fill price not strictly between 0 and 1: no fill. A signalled but unfilled market is not a bet; it is counted as n_unfilled_signals.

Model inputs.

• spot: the close of the last Alpaca 1-minute bar that ended at or before t_dec. Crypto uses BTC/USD, ETH/USD, SOL/USD; stocks use the SIP feed with raw (unadjusted) prices.
• sigma: annualized standard deviation (ddof=1) of daily log returns between the closes of Alpaca daily bars that started within the 30 days before t_dec and ended at or before t_dec (no partial bar); × √365 for crypto, × √252 for stocks.
• T: (end − t_dec) in days / 365.25.
• r = 0.05.
• p_model comes from the archived pricer: bs_prob_above (above), 1 − bs_prob_above (below), bs_prob_touch_above, bs_prob_touch_below.

Signal. edge = p_model − yes_price.

• Buy YES if edge > 0.08.
• Sell YES (buy NO) if edge < −0.08.
• Otherwise no bet.
• 0.08 is the archived bot's 6% edge plus its cost buffer. It is fixed a priori, not tuned on TRAIN.

Returns. Hold to resolution, $1 notional per bet, at p = fill_price.

• Buy at price p: (1 − p) / p if YES resolves, otherwise −1.
• Sell at price p: p / (1 − p) if NO resolves, otherwise −1.

Costs. oddsx.research.validation cost model on fill_price, spread 0.02.

• fee_rate from the market's fee flags: feesEnabled false → 0.0; crypto_fees_v2 → 0.07; finance_prices_fees or tech_fees → 0.04; any other enabled fee type → 0.05.

Split. Calendar split on t_dec: TRAIN before 2026-07-01T00:00:00Z, TEST on or after it. TRAIN is reported, never used to choose anything.

Benchmark. The favorite null on the filled decisions, same costs, at fill_price. Buy YES when yes_price ≥ 0.5, otherwise sell YES. edge_vs_naive = strategy TEST mean net − benchmark TEST mean net.

Verdict rule (TEST, pooled).

• INSUFFICIENT_DATA if TEST has fewer than 200 bets.
• VALIDATED only if all three hold: mean net > 0, edge_vs_naive > 0, and the 95% lower bound of mean net > 0. The lower bound comes from a cluster bootstrap: clusters are (symbol, end), 2000 resamples, seed 7.
• FALSIFIED otherwise.

Also reported (no effect on the verdict).

• Per-cohort results by asset type and side.
• Median vs mean net return.
• The top-1% |net PnL| share.

Promotion. A VALIDATED verdict does not move anything into strategies/. Promotion is a separate operator decision.

Results

Computed by research/hypotheses/fundamental_arb_exec/run.py --results on runs/g4c/decisions.parquet (commit 7542179). Numbers are copied from registry/results/fundamental-arb-exec.json and runs/g4c/universe.json.

Universe accounting. 106,318 markets parsed in the window; 29,623 kept (one decision each), the same universe as fundamental_arb_short. Exclusions:

Fill accounting. 4,421 of the 29,623 decisions have a fill trade in [t_dec, t_dec + 10 min). Fill delay after t_dec: median 198 s, p10 17 s, p90 492 s. Signals: TRAIN 1,546 (627 filled, 919 unfilled), TEST 1,228 (516 filled, 712 unfilled). These signal counts equal fundamental_arb_short's bet counts, as they should: the signal is unchanged.

TRAIN vs TEST (pooled, $1 per bet at fill_price, net of costs unless stated).

TEST cohorts (reported, no vote). crypto|above: 515 bets, mean net −0.2456. crypto|touch_above: 0 bets. stock|above: 1 bet, mean net +0.1202.

Tails. In TEST the median is slightly positive (+0.009) while the mean is −0.245: many small wins are outweighed by full −1 losses. The largest 1% of bets carry about 14% of absolute PnL, so the loss is not a few outliers. In TRAIN the mean is about three times the median (+0.113 vs +0.039) and the largest 1% carry about 27% of absolute PnL; even TRAIN barely beats the favorite null (+0.008) and its lower bound is negative.

Caveats.

• Limited test: realized volatility for every asset. The archived bot's ATM implied volatility for equities was not tested.
• TRAIN never votes.
• TEST is effectively one cohort (crypto|above, 515 of 516 bets), so the pooled verdict says nothing about stocks or touch markets.
• The fill is the first trade after t_dec, not an order-book quote. A signalled market with no trade in the next 10 minutes is dropped.

Post-hoc diagnostics (not part of the verdict)

Computed by the controller on 2026-09-14, after the verdict. None of this was pre-registered and none of it changes the verdict below. It bears on why the fundamental_arb_short edge disappears.

1. No selection bias from unfilled signals. TEST signals valued at the stale signal price (fundamental_arb_short economics): filled n = 516, mean net +0.179; unfilled n = 712, mean net +0.120. The unfilled signals were not the better ones, so dropping them does not explain the loss.
2. The price moves against the signal. Fill price minus signal price, in the bet's direction: median +0.044, mean +0.089. 67.4% of fills are worse than the signal price.
3. Same bets, two prices. The 516 filled TEST bets earn +0.179 net at the stale signal price and −0.245 net at the executable fill.
4. Fill speed does not save it. Mean net at signal price → at fill, by fill delay:

1. Conclusion for fundamental_arb_short (G4b). Its frozen-rule result (TEST mean net +0.145) is explained by stale YES prices. With executable entry the edge disappears.

Verdict

FALSIFIED

Under the pre-registered rule, TEST has 516 bets (above the 200 floor), and all three conditions fail: mean net −0.2449 ≤ 0, edge_vs_naive −0.1945 ≤ 0, and the cluster-bootstrap 95% lower bound −0.3901 ≤ 0. The protocol verdict is recorded as computed.

Plain conclusion: the short-horizon Black-Scholes signal has no executable edge on these markets. fundamental_arb_short's result was a stale-price artifact: the market price catches up to the model before a bet can be filled. Do not paper-trade or promote any fundamental_arb variant; nothing moves into strategies/.