Skip to main content

analyzed correlation between my strategies - found hidden risk

ran correlation analysis on my strategies.

found something i didn’t expect.

what i analyzed
#

all trades from may-july (3 months, 65 trades).

grouped by strategy type:

  • SPX options (25 trades)
  • TLT options (18 trades)
  • sector ETF options (12 trades)
  • IWM/QQQ options (10 trades)

calculated correlation between strategy returns.

the correlation matrix
#

learned about correlation analysis and ran it on my strategies:

             SPX    TLT   Sector  IWM/QQQ
SPX         1.00   0.15    0.72    0.68
TLT         0.15   1.00    0.08    0.12
Sector      0.72   0.08    1.00    0.65
IWM/QQQ     0.68   0.12    0.65    1.00

what this means
#

low correlation (good):

  • TLT vs everything else (0.08-0.15)
  • bonds move different than equities
  • provides real diversification

high correlation (problem):

  • SPX vs Sector ETFs (0.72)
  • SPX vs IWM/QQQ (0.68)
  • Sector vs IWM/QQQ (0.65)

basically: all my equity strategies are highly correlated.

the hidden risk
#

when equities dump, ALL my equity positions lose together.

not actually diversified.

example from july 10 (my red day):

  • SPX put spread: -$301
  • IWM put spread: -$336
  • both lost same day, same reason (equity dump)

TLT trade that day: +$93 (bonds rallied while stocks dropped)

what i’m changing
#

new rule: max 2 correlated positions at once

implementation:

if holding SPX position:

  • can add 1 more equity position (sector/IWM/QQQ)
  • can add unlimited TLT positions (uncorrelated)
  • cannot add 2nd SPX position

if holding 2 equity positions:

  • block new equity entries until one closes
  • still allow TLT entries

why TLT is key
#

TLT correlation to equities: 0.08-0.15

when stocks drop, bonds often rally (flight to safety).

TLT provides actual hedge.

backtesting the new rule
#

applied “max 2 correlated positions” to july trades.

july actual:

  • some days held 3-4 equity positions
  • correlation risk: high
  • july 10 red day: -$544 (multiple correlated losers)

july with new rule:

  • would’ve blocked 8 equity entries
  • missed: 5 winners (+$1,840), 3 losers (-$890)
  • net missed: +$950

but:

  • july 10 would’ve been -$301 instead of -$544 (blocked IWM entry)
  • reduced concentration risk significantly

tradeoff: give up ~$950/month in gains for lower max drawdown.

worth it for risk management.

implementation details
#

added to trade entry checklist:

async def can_enter_position(self, symbol: str) -> bool:
    """Check if position entry allowed"""

    # Get current positions
    equity_positions = self.get_equity_positions()

    # Check correlation
    if self.is_equity(symbol):
        if len(equity_positions) >= 2:
            logger.warning(f"BLOCKED: {symbol} - max equity positions (2) reached")
            return False

    # Other checks (vol, existing exposure, etc.)
    return True

monitoring
#

added correlation dashboard to grafana:

  • tracks current correlation between positions
  • alerts if correlation > 0.70 on new entry
  • shows diversification score

what this means for august
#

before: could hold 4-5 equity positions at once

now: max 2 equity positions at once

effect:

  • lower total positions
  • better diversification
  • reduced concentration risk
  • slightly lower expected return but way lower risk

the lesson
#

thought i was diversified (SPX, sectors, IWM, QQQ).

actually wasn’t. all moved together.

TLT is only true diversifier.

need to focus on uncorrelated strategies more.

next steps
#

short term: enforce max 2 equity positions

medium term: research more uncorrelated strategies

  • commodities (GLD, oil, nat gas)
  • international (EEM, EFA)
  • volatility (VXX)

long term: build truly diversified portfolio of strategies

trading update
#

wednesday 8/2: +$405 (TLT call spread)

thursday 8/3: +$280 (SPX put spread)

august total: +$1,430

goal: +$4,000. need: +$2,570 remaining.


2:18pm thursday. found hidden correlation risk. implementing max 2 equity positions. better risk management.

-AK

Related

correlation risk - learned the hard way
lost $1,400 yesterday because i didn’t track correlation between my positions. dumb mistake. what happened # may 10, 2pm:
testing tighter stops - early results mixed
started testing tighter stop losses today. goal: reduce avg loss from $660 to $600. early results: mixed. the change # old stops: -150% of credit received
analyzing profit factor - why 2.1 isn't good enough
been tracking profit factor for 2 months. current: 2.1 sounds good. but it’s not good enough. what is profit factor # profit factor = gross wins / gross losses
lost one, saved three - vol detection paying off
thursday was rough but could’ve been way worse. what happened # thursday 6/8: TLT put spread: -$315 (stopped during bond market dump) account: $341,720 → $341,405 lost $315 on one trade. sucks but manageable.
IB increased my margin - didn't ask for it
woke up to email from interactive brokers. “Your account has been approved for portfolio margin” didn’t apply for it. they just gave it to me. what is portfolio margin # reg-t margin (what i have now):
may day 1 - smaller position sizes
cut position sizes in half. only trading 3 strategies now. first day of may went… fine? +$180 on one trade. nothing else triggered. new risk rules # old position sizing: 2.5% risk per trade = $850 max loss (on $340k account)