7. Smart money leaving
Should a position be sold when the holders behind it leave, and should money that fell before the signal block the entry?
Definition
Exits on the baseline's entries: sell when the biggest holder at entry has cut 50 %+ of its shares, when the side's money fell 50 %+ from its peak since entry, or either. Entry filter at signal: the side's money fell 25 %+ over the previous 6 hours (markets younger than 6 h count as not fallen).
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 |
|---|---|---|---|---|---|---|---|---|
| 7a exit when the biggest holder at entry cut >= 50 % strategy, compared with base-hold | unknown | −4.7 % [−8.1, −1.4] n = 3,064, hit 59 %, price 0.62 | −5.1 % [−7.3, −2.8] n = 3,736, hit 63 %, price 0.64 | −4.9 % [−6.9, −3.0] n = 6,800, hit 61 %, price 0.63 | IS −3.8 % (3,564) OOS −5.5 % (2,743) | −0.5 [−1.3, +0.1] points vs base-hold, p = 0.9196 | no | negative |
| 7b exit when the side's money fell >= 50 % from its peak strategy, compared with base-hold | unknown | −4.7 % [−7.8, −1.4] n = 3,075, hit 58 %, price 0.62 | −5.5 % [−7.7, −3.3] n = 3,738, hit 61 %, price 0.64 | −5.1 % [−7.0, −3.2] n = 6,813, hit 60 %, price 0.63 | IS −3.4 % (3,565) OOS −6.6 % (2,745) | −0.6 [−1.6, +0.3] points vs base-hold, p = 0.8936 | no | negative |
| 7c exit on either strategy, compared with base-hold | unknown | −5.3 % [−8.2, −2.3] n = 3,076, hit 57 %, price 0.62 | −5.6 % [−7.7, −3.3] n = 3,727, hit 60 %, price 0.64 | −5.4 % [−7.2, −3.6] n = 6,803, hit 58 %, price 0.63 | IS −4.1 % (3,554) OOS −6.4 % (2,746) | −0.9 [−1.9, +0.2] points vs base-hold, p = 0.9545 | no | negative |
| 7d entry filter: the side's money did NOT fall >= 25 % in the last 6 h entry filter, decided at the signal | better | −3.9 % [−7.3, −0.2] n = 3,043, hit 64 %, price 0.62 | −4.9 % [−7.3, −2.4] n = 3,733, hit 65 %, price 0.64 | −4.5 % [−6.6, −2.4] n = 6,776, hit 64 %, price 0.63 | IS −2.8 % (3,564) OOS −6.1 % (2,720) | −20.9 [−70.4, +20.1] points vs the dropped trades, p = 0.8127 | no | negative |
| 7d-fell the side's money fell >= 25 % in the last 6 h (the dropped set) entry filter, decided at the signal | worse | n = 20 (under 30) | n = 23 (under 30) | +16.5 % [−25.0, +62.1] n = 43, hit 60 %, price 0.57 | IS +20.4 % (18) OOS +24.0 % (22) | +20.9 [−20.5, +73.8] points vs the dropped trades, p = 0.1863 | no | nothing |
What this data can and cannot show
Can: Whether these exits lose less than the lost / flip exits of the core study did, and whether a draining side is a worse entry.
Cannot: Sales outside the tracked set, and (May-October 2026) merges, which the activity export lacks: some exits are invisible and the holder stays in place.
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.