3. Buying below smart money's entry
Does buying the consensus side cheaper than the holders paid beat buying it after the price moved their way?
Definition
The volume-weighted entry price of the consensus side's holders (sum of shares x average price / sum of shares) minus the side's current price at the hour the consensus formed. Variants: 5+ points below the holders' entry, 10+ below, within 5, 5+ above.
Baseline for every comparison: the plain consensus on the same markets (5+ tracked holders, 60 %+ on one side, price 0.05–0.95, first signal, held to resolution), 1 ¢ half-spread plus Polymarket's taker fee; $1 a trade.
Mean return per trade, with 95 % intervals
One marker per window: circle = A (curated wallets, Mar–Apr 2026, clean), square = B (all tracked wallets, Oct 2026, provisional, drift ranges dropped), diamond = pooled. Variants with fewer than 30 trades in a window are left out of the chart; the table below has every number.
Every variant
| Variant | Expected | Window A | Window B | Pooled A + B | IS / OOS (pooled) | vs its baseline (pooled) | Guard | Verdict |
|---|---|---|---|---|---|---|---|---|
| 3a price >= 5 points below the holders' entry entry filter, decided at the signal | better | −14.7 % [−28.4, −0.9] n = 450, hit 41 %, price 0.40 | −15.4 % [−24.8, −7.2] n = 460, hit 50 %, price 0.53 | −15.1 % [−23.2, −6.7] n = 910, hit 46 %, price 0.46 | IS −20.8 % (514) OOS −9.7 % (324) | −12.4 [−21.2, −3.9] points vs the dropped trades, p = 0.9965 | no | negative |
| 3b price >= 10 points below the holders' entry entry filter, decided at the signal | better | −24.9 % [−41.1, −8.5] n = 287, hit 33 %, price 0.32 | −14.6 % [−28.4, +1.2] n = 222, hit 43 %, price 0.44 | −20.4 % [−30.7, −8.8] n = 509, hit 37 %, price 0.37 | IS −25.6 % (309) OOS −14.0 % (164) | −17.3 [−28.1, −5.3] points vs the dropped trades, p = 0.9935 | no | negative |
| 3c within 5 points of the holders' entry (control) entry filter, decided at the signal | control | −3.2 % [−7.8, +1.6] n = 1,845, hit 62 %, price 0.61 | −3.9 % [−7.1, −0.9] n = 2,630, hit 64 %, price 0.63 | −3.6 % [−6.3, −0.9] n = 4,475, hit 63 %, price 0.62 | IS −1.2 % (2,217) OOS −5.5 % (1,948) | +2.1 [−2.6, +6.6] points vs the dropped trades, p = 0.1833 | no | nothing |
| 3d price >= 5 points above the holders' entry (control) entry filter, decided at the signal | worse | +1.5 % [−3.3, +6.3] n = 768, hit 81 %, price 0.79 | −1.2 % [−5.4, +3.1] n = 666, hit 78 %, price 0.77 | +0.2 % [−2.8, +3.3] n = 1,434, hit 80 %, price 0.78 | IS +4.6 % (851) OOS −4.7 % (470) | +5.8 [+1.9, +10.0] points vs the dropped trades, p = 0.0045 | no | nothing |
What this data can and cannot show
Can: Whether the holders' cost basis carries information about the price, or whether a price that fell after they bought means they were wrong.
Cannot: The order book: the price is the last tracked fill (median error 0.5 c, p90 2.5-16 c), so small differences are noise; the 5-point bar is there for that reason.
Caveats that apply to every signal
- Only tracked wallets' fills exist, from when the site started tracking them; the replay is closer to the chain than to what the site displayed (no Polymarket position overrides, merges missing May–Oct 2026).
- Window B's fill times run early in parts of October 2026; the affected ranges are dropped and the rest is not proven clean. Its numbers are provisional until the repaired export is in.
- Prices are the last tracked fill, not the order book (median error 0.5 ¢, p90 up to 16 ¢ on fast markets); entries happen on the hour (5-minute steps in play).
- The windows are 7 and 3 weeks long; effects of a few points cannot pass a guard over 40 variants there. A failed guard is not evidence of no effect — it is the absence of evidence.
- Nothing here is trading advice. Past replays, even clean ones, do not pay future trades.