@quantpad: Stop loss slippage backtest: the Swiss franc crash on the CME tape and the 100 worst stop fills on 8 futures markets since 2010 On January 15 2015 the franc jumped 25.4% in 25 minutes. FXCM's clients owed it $225 million and Alpari UK went insolvent. So did a stop fill that morning? I read every trade of 6SH5 off the CME tape: a short from 0.9797 with a buy stop one 14-day average range away filled at 0.9876, 1% worse than planned, $988 a contract. Fifteen ranges away, 3.4% worse. Without a stop it was $31,063 at the high and $19,900 at 12:00 UTC. The one real hole was 11 cents in 2 minutes with no trade, which cost a 20-range stop 1.49x its plan. So when does a stop fail? When the market is shut. Of the 100 worst misses on ES, NQ, CL, GC, ZN, 6E, 6J and 6S from 2010 to 2026, 39 came at a weekend reopen, 36 overnight, 5 at the 18:00 reopen and 20 in regular hours. One gap can count at several stop widths. Crude, Sunday March 1 2026: stop 68.00, Friday close 67.29, filled 75.00, 4.62x the plan. At a one-range stop, 1,165 S&P and Nasdaq stops entered at 9:30 and flat by 4 never gapped past on one-minute bars (held 5 sessions: 2.00x and 1.46x). Not everywhere: the yen and franc futures were worse day-only (3.41x, 1.99x), so this is not a rule for thin hours. Walked forward, expanding by year from 2012 with a one-session embargo, planning for the worst of all 8 markets so far: held trades beat the plan in 8 of 120 market-years (worst 1.43x), flat by the close in 0 of 120. Own market only: 18 and 7. Gaps cost held crude about $16 a stop, against a $16.50 round turn. Not modelled: order size and the queue (first trade past the stop, one contract), stop-limit protection, retail FX execution, a different market. Measurement, not a strategy. I built this in QuantPad. It has every trade on the CME since 2010. Link in bio. #QuantPad #stoploss #futurestrading #backtesting #riskmanagement

1391

There are no more comments for this video.

To see more videos from user @quantpad, please go to the Tikwm homepage.