News spikes: spreads blow out and books thin.
If your strategy trades through scheduled events, stress with wider slips on those windows or hard no-trade flags.
Session boundaries matter for futures.
A signal at 15:59 that fills next open is a different trade than a same-session fill. State the session clock you use.
Multiple testing without a budget is how every chart pattern "works".
Decide your search space before you look. Cap how many variants you try. Report the failures too.
Cancel/replace behavior matters for working orders.
A backtest that assumes resting limits sit forever will overstate fills versus a desk that pulls on adverse moves.
Scale-in and scale-out need explicit triggers.
"Add on strength" without size caps and add levels is not a rule. Write max adds, spacing, and what cancels the plan.
Intrabar assumptions matter. Using the high as an exit you "would have caught" is a common cheat.
Be explicit: fill on close, next open, or stop with gap risk. Fake fills inflate edges.
In public I will show diagrams labeled illustrative.
If I did not run it, I will not paste numbers. That rule keeps the feed honest when engagement wants fireworks.
Timing: signal on close, fill on next open is a different strategy than same-bar fill.
Pick one. Mixing them in research and live is how "mysterious" slippage appears.
Separate ideation time from evaluation time.
When you are still inventing filters, you are not evaluating. When you evaluate, freeze the code and only read scores.
Filters stacked until the trade count collapses are usually curve fit armor.
If you need five conditions to make the curve pretty, ask whether any one filter was pre-specified or added after looking.
Gross expectancy that ignores fees is a story, not a result.
Re-run with round-turn costs at a pessimistic level. If the edge vanishes, you never had one you could take live.
Partial fills happen when size exceeds depth.
Size your research to a depth you can actually take. Scaling a 1-lot edge to 50 lots without a fill model is fantasy.
Failure modes are research output.
"Does not survive 2-tick slip" is useful. Archive it with a note. That saves you from rediscovering the same dead edge next quarter.
Same-bar entry and exit assumptions will inflate results on noisy bars.
Decide bar resolution first. Then decide whether stops are checked before targets. Order of operations changes P&L shape.
Walk-forward is not magic. If each window re-optimizes freely, you still snoop.
Freeze the process. Same search rules every fold. Then check that the rule set survives, not just one lucky fold.
57 Followers 101 FollowingAlgotrader | 8+ years backtesting and trading | Stocks, ETFs & Futures | Amibroker.
Building systems that survive live trading. Sharing what actually works.
97K Followers 1K FollowingCo-Founder @SMBCapital and https://t.co/SF1XnbDLEy
Author- One Good Trade and The PlayBook
Board member- @Traders4ACause and @Impact_Comp
111K Followers 35K FollowingEst. 2015. Markets, speculation, risk. Podcast hosted by Kevin Avery. 25M+ plays.
For inquiries, reach out to [email protected]
14K Followers 433 FollowingWe inspire, empower & educate a global community of quants, providing world-class infrastructure for them to design and trade investment strategies.
30K Followers 5K FollowingPortfolio manager with relative strength and vol strategies. Ask me about PPLI. Also hack guitar player who sings. Not investment advice. Do your own work!!
87K Followers 877 FollowingCo-CIO of Ambrus| (Focus: Volatility Trading / Tail risk hedging )| @penn guy ( These are my personal thoughts and not the opinions of Ambrus)
96K Followers 973 FollowingPresident of @Spectra_Markets. Author of am/FX and Alpha Trader. I like trading, table tennis, my family, writing, hard/fast music, and poker. Not in that order
579K Followers 475 Followinghttps://t.co/SgaSuGdrox macro & beta @2Graybeards for beta. Both for investor education, Brevan Howard, Bridgewater, Salomon, Dad of 4. Go Penn, No tweet is advice
32K Followers 428 Followingvol trader, opinions are my own and may or may not be reflective of positions i hold. nothing should be taken as financial advice.
16K Followers 153 FollowingTrend, value, quality, carry, volatility. Tentmaker (eating vol-targeted Sortino to fund Christian work). See the pinned tweet for compound return generators.
15K Followers 671 FollowingHi...My name is Niels Kaastrup-Larsen. I've been part of the hedge fund industry for more than 35 years & since 2014 have helped democratise it. Join me!
80K Followers 427 FollowingAlgorithmic trader | 7 years live | 10+ algos live
Turning every trader algorithmic.
Co-founder @AskEdgarIO (agents for funds).
Advisor to Long Ridge EP.