Which Trading Psychology Concepts Actually Work? (Tested With Claude)

Summary

Brendan describes using Claude to test 83 popular trading-psychology concepts in a simulated trading system. The results suggest that many rules depend on whether a trader is profitable, how often they trade, and whether wins or losses cluster together (“seriality”).

The presenter’s central claim is that discipline cannot turn an unprofitable strategy into a profitable one; it mainly changes the pace and risk of results. Position sizing is presented as an important exception.

Testing framework and assumptions

• Claude applied testable rules to identical sequences of trades and compared each result with a control.
• The model used a profitable baseline trader with a $50,000 account, risking $500 per trade (1%), winning about 37% of trades, and making approximately 750 trades per year—about three per trading day.
• The same year of trades was simulated 10,000 times to account for variation between years.
• The model also tested seriality: whether a loss makes subsequent trades more likely to lose, or a win makes subsequent trades more likely to win.
• These are simulated, assumption-dependent results, not a report of live-trading performance.

Main findings

Losing streaks and sample size

• Losing streaks are normal: The baseline trader had a 94% chance of experiencing at least 10 losses in a row in a year; the median longest streak was 12 losses. The chance was 99% for a trader with no edge and 54% for an “expert” described as 3.5 times better than the baseline.
• A small sample tells little: After 100 trades, a profitable strategy still had a 24.5% chance of showing a loss. The presenter says roughly 800 trades are needed to get a clearer view of a strategy’s true edge; after just 10 trades, the sample provides little information.

Responding to losses and wins

• Revenge trading and doubling after losses: In the simulation, doubling position size after each loss raised the median ending account balance from about $97,000 to $157,000. But average position size became roughly 3.5 times larger, so the higher return reflected greater exposure—not a benefit from trading emotionally. The approach also increased the chance of losing half the account by 2 percentage points, with repeated doubling driving ruin risk higher.
• Hesitating after a loss: Skipping the next trade after two consecutive losses cost about $17,000 and meant missing roughly 28% of the year in the model. Stopping for the entire day after a substantial loss meant missing about half the year and cost approximately $30,000. Results depend on seriality: skipping after losses may help if losses cluster, but hurt if they do not.
• Stopping after two losses (“tilt” rule): For a trader without a tendency toward losing streaks, this rule cost about $8,442 per year. With a streak tendency, it added about $14,721 by avoiding trades that were worse than average. The rule began to help in the model at around four trades per day, with greater benefits at higher trading frequencies.
• Daily profit targets: Stopping after reaching a profit target cost a profitable trader about $15,000 per year in the simulation, with a larger cost when wins clustered. Chasing a target while in a loss was estimated to cost about $3,000 in the worst case tested. The presenter argues that fixed targets can limit a profitable system’s opportunity, while acknowledging that outcomes depend on the trader and system.
• Increasing size after winning streaks: Increasing position size after two, three, or four wins—or after reaching a daily, weekly, monthly, or account-equity high—showed no measurable advantage over taking the same larger trades at other times.

Selectivity and position sizing

• Selectivity and setup quality: Taking only above-average setups produced about 375 trades per year and cost approximately $85 in the model. Becoming very selective—taking only the top 17% of setups—generally reduced results. Instead, using setup quality to size up stronger trades generated about $32,000 per year, versus roughly no gain from skipping weaker trades, assuming low trading costs. At three times typical commissions and slippage, the estimated cost of selectivity rose to $12,575.
• Sizing up the best setups: The simulation’s estimated break-even point was about 1.5× normal size on stronger setups. The presenter cautions against moving to 2× or 3× unless an individual trader’s own data supports it. The video also notes that a 20% account loss requires a 25% gain to recover.

Capital protection and tail risk

The model tested a trade losing 20 times the planned amount because of a market gap through a stop. Under that scenario, the annual chance of losing half the account was:

• 17% at 1% risk per trade
• 56% at 2% risk per trade
• About 99.9% at 5% risk per trade

The presenter says the Kelly criterion indicated 5% for the modeled trader, but warns that Kelly sizing requires knowing one’s true edge. He presents 1% risk per trade as a more cautious choice when skill is uncertain.

Practical takeaways and tools

• For traders who are not yet profitable, the presenter says to focus on fixing the trading system; rules that reduce trading may limit losses but do not create an edge.
• For profitable traders, the video favors taking more trades and increasing size selectively on stronger setups, while limiting increases to around 1.5× unless personal results justify more.
• Skipping a trade after a loss may be useful for profitable traders whose losses demonstrably cluster. The presenter recommends testing this rather than assuming it applies.
• The video proposes four Claude analyses using a broker-exported CSV:
  1. A seriality check.
  2. A day-trader comparison of trading limits.
  3. An assessment of A+ setups and position sizing.
  4. A survival check for outsized losses.

The seriality check is said to require at least one year of history, with about two years needed for results described as accurate roughly 19 times out of 20.

Assets, instruments, and disclosures

• Tickers, securities, sectors, and specific instruments: None mentioned.
• Disclosure: Brendan states that the video is not financial advice and does not guarantee profits.

Presenter and sources

• Presenter: Brendan says he studied mathematics and economics at the University of California, worked in investment banking for over three years, and has spent the last two years creating trading systems for clients.
• Source/tool: Claude was used to build and run the simulations. The concepts were gathered from popular trading-psychology videos.