5 Discipline Mistakes That Cost Me $20K (Case Study)

The $20K Wipeout Nobody Warned Me About

In October 2023, I lost $20,387 trading ES futures in fourteen calendar days. Not from a blown risk model. Not from a black swan. From five discipline failures that every trading psychology book mentions—but that I convinced myself didn't apply to me.

I'm a software engineer who passed a $50K Apex evaluation in August 2023. My edge was simple: I trade the 9:30 AM EST open using a volume profile and order flow setup that wins 58% of the time over 200+ trades. My average winner ran 12 ticks; average loser stopped at 8. Clean 1.5:1 risk-reward, tight stops, boring execution.

Then I violated my own rules five times in two weeks. This is the breakdown.

Mistake 1: Doubling Position Size After a Win Streak (Recency Bias)

Days 1-3: The setup

I entered October on a seven-win streak—my best run in three months. Each trade risked $400 (4 ES contracts, 8-tick stop). I'd banked $3,360 in five trading days. My inner dialogue: "I'm seeing the market clearer than usual. This is when you press your edge."

On October 4th, I doubled to 8 contracts without adjusting my stop. Same setup, same 8-tick stop, now risking $800.

The math:

A single loss now erased two average winners instead of 0.8 winners. My Kelly Criterion optimal position—given my 58% win rate and 1.5:1 R/R—was roughly 24% of equity per trade. I'd jumped to 48%.

The result:

I hit three losses in the next four trades. Those three losses (-$2,400) erased seven wins of profit and added a $960 hole. Kahneman and Tversky's work on recency bias (from Thinking, Fast and Slow) describes exactly this: recent outcomes disproportionately weight our probability estimates. Seven wins convinced me I was "in the zone," not that I was overdue for mean reversion.

Mistake 2: Revenge Trading After a Bad Fill (Loss Aversion)

Day 5: The trigger

October 9th. I placed a limit order at 4412.75 on ES. Market ticked to 4412.50, triggered stops above, and ripped to 4415 without filling me. Classic. I watched my setup play out for 12 ticks—$600 per contract I "should have made."

Instead of waiting for the next signal, I market-bought at 4415.25. No volume profile confirmation. No order flow. Just anger.

The rationale I told myself:

"The market owes me this trade." Textbook loss aversion from prospect theory—the pain of missing a winner hurt more than the risk of a new loser.

The damage:

Stopped out 11 ticks later for -$880 (8 contracts, still oversized from Mistake 1). The original setup I missed would have been a 4-contract trade risking $400. This revenge trade cost 2.2x a normal loss on a signal that didn't exist.

Traders who read about cognitive biases recognize this instantly. I'd read about it. I still did it.

Mistake 3: Moving My Stop Loss Mid-Trade (Commitment Bias)

Days 6-8: The slippery slope

October 10th. I took a valid short at 4421.00, stop at 4429 (8 ticks). ES chopped and tagged 4428.75—one tick from my stop—then reversed 6 ticks in my favor.

"See? It would've worked. My stop is too tight."

October 11th. Same setup, different price. I placed my stop, then moved it 3 ticks wider mid-trade when price approached. Got stopped out at the new level. The original stop would've held, and I'd have caught an 11-tick winner.

The psychology:

I committed to the trade idea but not the risk parameter. Mark Douglas describes this in Trading in the Zone as "hoping" instead of executing. Moving a stop mid-trade is statistically just taking a larger loss—but it feels like control.

The cost:

Three trades, three moved stops, three losses totaling $3,120. If I'd kept original stops: one loss, two scratch trades. Actual cost of this bias: ~$2,500.

Mistake 4: Overtrading to "Make It Back" (Sunk Cost Fallacy)

Days 9-11: The spiral

By October 12th, I was down $6,400. I started taking B-grade setups—trades that met 80% of my criteria but not 100%. My logic: "I need more at-bats to recover."

I took 23 trades over three days. My normal baseline is 2-3 per day. Win rate collapsed to 39% because I was forcing entries.

The math on overtrading:

• Normal: 3 trades/day × 58% win rate × 1.5 R/R = +0.61R per day
• Desperate: 7.7 trades/day × 39% win rate × 1.5 R/R = -0.45R per day

I lost $4,180 in this phase. Not because my edge vanished—because I abandoned it to recover sunk costs faster.

If you've explored the Risk Management category on this site, you've seen this pattern: drawdowns compound when you change behavior because of the drawdown.

Mistake 5: Ignoring My Own Stop-Out Rule (Confirmation Bias)

Days 12-14: The final push

My personal rule: stop trading for the day after two consecutive losses. It's in my plan. I've never regretted following it.

October 16th. Two losses before 10:00 AM. I told myself, "The third setup is perfect. I can't miss this."

It wasn't perfect. I wanted it to be perfect because I needed a win. Confirmation bias—I saw order flow that confirmed my directional bias and ignored the volume profile showing weak internals.

Third loss. Then a fourth, because "I'm already down, might as well see if the afternoon session gives something back."

Final tally:

Four trades, four losses, $3,200 gone. If I'd stopped after two, I'd have saved $1,600 and preserved mental capital for October 17th, which delivered two clean winners I was too tilted to take.

The Intervention: What Actually Changed

I didn't fix this with willpower. I rebuilt my process:

1. Position size cap in writing: Printed a sheet with max contracts per trade (4 ES) and taped it to my monitor. Childish? Maybe. Effective? Yes.

2. Mandatory journaling: I started logging emotional state before每 trade using NinjaTrader's trade notes field. If I typed "need to make back losses" or "market owes me," I closed the platform.

3. Hard stop-loss locks: I use Tradovate's bracket orders with a stop that cannot be modified after entry. If I want to move it, I have to flatten and re-enter—friction that makes me reconsider.

4. Daily loss limit: After -$800 in a day, I'm done. No exceptions. I automated this with a P&L tracker that emails me when I'm within $100 of the limit.

5. Real-time bias detection: I tested a Chrome extension (MindGuard) that flags potential cognitive biases while trading on Tradovate. It caught three "oversized position" warnings in November that I would've ignored. Not a silver bullet, but another friction layer. You can see how it works on the Features page.

The Outcome: 90 Days Later

From October 17 to January 15, I traded the same strategy with the rebuilt process:

• 127 trades
• 56.7% win rate (slightly below baseline—regression to mean after the hot streak)
• Average R per trade: +0.48
• Largest single loss: $560 (1.4R—violated my stop once in three months)
• Total P&L: +$6,140

I didn't recover the $20K. I rebuilt consistency. The account is still down $14K from peak, but I'm executing my edge again instead of fighting my psychology.

The five trading mistakes I made aren't novel. Tharp, Steenbarger, Douglas—they all document these patterns. But reading about discipline failure and experiencing a $20K account drawdown are different feedback loops. I had to lose real money to treat my trading rules like engineering constraints instead of suggestions.

If you're in a drawdown, the fix isn't motivation. It's friction: systems that make it harder to violate your process than to follow it.