Skip to main content

week 3 - small wins add up

first positive week in may. finally.

the numbers
#

trades this week:

  • total: 5
  • wins: 4
  • losses: 1
  • win rate: 80%
  • avg winner: $295
  • avg loser: $240
  • net: +$940

account:

  • started week: $339,510
  • current: $340,450
  • weekly return: +0.28%
  • may total: +0.14% (+$450)

back to green for the month.

what changed
#

simplified strategies:

spent monday checking for overfitting. found it everywhere.

cut rules from 12 to 5. removed oddly specific parameters.

backtests look worse but live performance better.

old strategy:

  • backtest sharpe: 2.08
  • live sharpe: 0.43
  • 79% degradation

new strategy:

  • backtest sharpe: 1.71
  • live sharpe: 1.52 (estimated from this week)
  • 11% degradation

way more consistent.

correlation limits working:

new code rejected 2 trades this week:

  • GOOGL put spread (0.87 correlation with existing QQQ position)
  • NVDA call spread (0.91 correlation with AAPL position)

would’ve both lost money on thursday’s tech selloff.

saved ~$600 by not taking them.

async market data:

rebuilt data pipeline using python async.

latency: 650ms → 120ms per update throughput: 10 symbols/sec → 80 symbols/sec

faster data = better entry prices = higher win rate.

no relationship distractions:

broke up with J on thursday. been single since.

weirdly focused now. no texts to respond to, no “where is this going” conversations.

just trading, coding, gym, sleep.

kind of prefer it this way.

week 3 trades
#

date underlying type credit outcome P&L reason
5/15 SPX put spread $0.92 win 50% $410 normal exit
5/16 TLT call spread $0.68 win 50% $295 normal exit
5/17 GLD put spread $0.55 loss -$240 stopped out
5/19 SPX put spread $0.85 win 50% $380 normal exit
5/20 IWM call spread $0.72 win 2DTE $295 early exit

note: all different underlyings, low correlation between positions.

may progress
#

week 1: +$520 week 2: -$1,010 week 3: +$940

net may: +$450 (0.14%)

finally positive for the month.

what’s working
#

1. position diversity

this week’s positions:

  • SPX (broad market)
  • TLT (bonds, -0.45 correlation to stocks)
  • GLD (gold, 0.12 correlation to stocks)
  • IWM (small caps, 0.76 correlation to SPX but different sector exposure)

max correlation between any two: 0.76

much better than last week’s 0.92.

2. simpler rules

premium selling strategy now:

  • IV rank > 45
  • delta 0.15-0.25
  • dte 30-45
  • earnings > 7 days away
  • correlation < 0.7 with existing positions

5 rules. clean. works.

3. smaller position sizes

still at 1.25% risk per trade ($425 max loss)

one loss this week cost $240. manageable.

4. realistic expectations

not targeting massive returns. just consistent small wins.

$940 per week = $4k per month = $48k per year

on $340k account = 14% annual return

not amazing but sustainable.

may goal status
#

original goal: break-even to +$5k

current: +$450

one week left. need +$4,550 to hit upper goal.

unlikely but at least i’m positive now.

the one loss
#

GLD put spread on 5/17.

entered at $0.55 credit, stopped out at -200% for $240 loss.

gold spiked on dollar weakness. nothing wrong with trade, just got stopped.

win rate will never be 100%. managed the loss properly.

correlation check prevented losses
#

thursday tech dump:

  • QQQ: -2.1%
  • AAPL: -2.4%
  • GOOGL: -2.8%
  • NVDA: -3.1%

had existing QQQ position. correlation checking rejected GOOGL and NVDA trades.

if i’d taken them:

  • QQQ position: -$415 (stopped out)
  • GOOGL position: -$485 (would’ve stopped)
  • NVDA position: -$520 (would’ve stopped)

total loss: -$1,420

actual loss: -$415 (only QQQ)

correlation checking saved $1,005.

psychological state
#

better this week.

focused. not distracted by relationship stuff.

trading is clearer when it’s the only thing i’m thinking about.

still miss my parents randomly. been 21 weeks. still sucks.

but trading gives structure. concrete problems with concrete solutions.

changes for week 4 (final week of may)
#

trading:

  • continue current approach (working)
  • stay strict on correlation limits
  • no changes to working strategies

code:

  • finish async order execution
  • add more backtest validation
  • document everything

personal:

  • gym 4x this week (been slacking)
  • therapy wednesday
  • no dating for a while

lessons from may so far
#

  1. simplicity beats complexity - fewer rules, better results
  2. correlation matters - saved $1k+ this week by checking it
  3. small consistent wins - better than big inconsistent wins
  4. overfitting is real - impressive backtests don’t mean shit
  5. focus matters - no relationship = more focus = better trading

may final week prediction
#

conservative: +$500 to +$1,000 (total: +$950 to +$1,450) realistic: +$800 (total: +$1,250) optimistic: +$1,500 (total: +$1,950)

would be happy with anything above +$1k for the month.

first profitable month since going live would be huge.

comparison to april
#

april: -$12,384 may (so far): +$450

massive improvement.

if may ends positive, proves strategies are fixable.

the bigger picture
#

started with $400k in january. lost $60k in first 4 months learning. currently at $340k.

if may ends at +$1k:

  • 5 month total: -$59k
  • burn rate slowing
  • might actually become profitable

long way to go but week 3 feels like turning point.


4:15pm sunday. week 3 done. one more week to finish may positive. feels possible now.

-AK

Related

week 1 review - small progress
first week of may done. +$520. not amazing but way better than april’s weekly average (-$3k/week). the numbers # trades this week:
week 2 review - back in the red
week 2 sucked. gave back week 1 gains and then some. the numbers # trades this week: total: 7 wins: 3 losses: 4 win rate: 43% avg winner: $310 avg loser: $485 net: -$1,010 account:
april wrap - down another 12k
april P&L: -$12,384 4-month total: -$60k not improving fast enough. the numbers # Figure 1: Monthly and cumulative P&L Jan-Apr 2023. Trend is slightly improving but still deep red.
first 90 days live - reality check
3 months into trading with real money and i’m down $40k. not what i expected but also exactly what i should’ve expected. paper trading for 3 years (2020-2022) gave me this false confidence that live trading would be similar. LMAO nope. completely different game when actual money is on the line.
march 2023: first month live trading results
first month done # march 15 - march 31 (2.5 weeks live) went from paper trading to real money. lost $180k in february learning. figured some shit out. went live march 15
week 3 performance: sharpe 2.1 actual vs 1.64 backtest
three weeks live = enough data for initial validation # march 15 - march 30 (15 trading days)