Skip to main content

testing momentum overlay - early results promising

been running mean reversion strategies for 4 months.

works well. 87% win rate in july.

but: only trades when price reverts.

misses trending moves.

the idea
#

add momentum overlay to existing system.

mean reversion: buy dips, sell rips, expect reversion to mean

momentum: follow trends, ride winners, cut losers fast

combine both: trade mean reversion in ranging markets, momentum in trending markets.

how to detect regime
#

ranging market indicators:

  • 20-day ATR < 50-day ATR (volatility contracting)
  • price within ±1 std dev of 20-day SMA
  • no new 20-day highs/lows in past 5 days

trending market indicators:

  • 20-day ATR > 50-day ATR (volatility expanding)
  • price outside ±1 std dev of 20-day SMA
  • new 20-day high/low in past 3 days

strategy logic
#

if ranging_market:
    use mean_reversion_strategy()
    # current system, already working

elif trending_market:
    use momentum_strategy()
    # new addition

else:
    wait_for_clear_regime()
    # skip unclear regimes

momentum strategy rules
#

entries:

  • buy breakout above 20-day high
  • sell breakdown below 20-day low
  • only in direction of trend

exits:

  • trail stop at 2 × ATR
  • profit target at 3 × ATR (risk:reward = 1:1.5)
  • max hold: 5 days

position size:

  • same 1.25% risk as mean reversion
  • but faster stops (2 ATR vs fixed dollar)

paper trading results (7 days)
#

ran in paper mode alongside live mean reversion.

trades: 4 momentum trades (all paper)

  • 3 wins: +$280, +$340, +$405
  • 1 loss: -$270

win rate: 75% (3/4) avg win: +$342 avg loss: -$270 profit factor: 3.8

vs mean reversion same period:

  • mean reversion: 6 trades, 100% win rate, +$2,065
  • momentum: 4 trades, 75% win rate, +$755 (paper)

key differences
#

mean reversion:

  • high win rate (80-90%)
  • small consistent wins
  • works in ranging markets
  • most of the time (70% of days)

momentum:

  • lower win rate (70-75% expected)
  • larger wins when right
  • works in trending markets
  • less frequent (30% of days)

combined: capture both regimes.

what i’m testing this week
#

running momentum strategy in small size (0.5% risk).

live money but reduced risk while validating.

criteria to go full size:

  • 20+ trades
  • win rate > 70%
  • profit factor > 2.0
  • no correlation increase with mean reversion

potential issues
#

1. regime detection lag

by the time i detect trend, might be late.

solution: use faster indicators (10-day instead of 20-day).

2. whipsaws

fake breakouts trigger entries, then reverse.

solution: require volume confirmation on breakouts.

3. correlation with mean reversion

if both strategies lose together, not adding diversification.

solution: track correlation daily, adjust if >0.5.

trading update
#

monday 8/7:

  • mean reversion: GLD put spread +$385
  • momentum (paper): none (no trending regime detected)

august total: +$2,450

goal: +$4,000. need: +$1,550 in 3 weeks.

next steps
#

continue paper trading momentum for 2 more weeks.

collect 20+ trades.

then decide: add to live system or abandon.


3:55am monday. testing momentum overlay. early paper results promising. might add regime diversification.

-AK

Related

ran walk-forward test - strategy holding up
been profitable 3 months now. wanted to verify i’m not just curve fitting. ran walk-forward analysis this weekend. what is walk-forward testing # normal backtest:
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
vol detection system is live
went live with vol regime detection friday morning. first real test today. what happened # saturday market (crypto):
looking ahead to june
may is wrapped. planning june now. what i’m carrying forward # strategies: premium selling (working) mean reversion v2 (needs more testing) vol regime detection (deploying monday) risk management:
building volatility regime detection
need to stop trading when volatility spikes. building detection system. the problem # this week VIX spiked 18% in 2 days. my strategies got stopped out twice.
checking my backtests for overfitting
worried my strategies are overfit to historical data. spent today testing for it. been reading NexusFi backtesting threads about this exact problem. the problem # my backtests look great: