A few weeks ago I shipped V5 after returning from my work trip to Austin. Tuned VAMS parameters, Arsenal SPY veto, Sharpe at 0.991, max drawdown at -14.3%. The engine that finally crossed the threshold from “good” to “real.”
I was happy with V5. The plan was to live with it, watch it work, maybe tinker on Arsenal sleeves.
That plan lasted about a week.
There it was.
V6 had a slightly better worst-case drawdown.. but 3.4 percentage points more of its life was spent in -5% to -10% drawdown territory.
That’s roughly 200 extra days underwater at -5%+ over the sample.
The headline number lied. Or rather, it told the truth about ONE moment (the worst single drawdown), while hiding the truth about every other moment.
My eyes were right. The chart was heavier.
The mechanism was obvious in hindsight: half-lifts during MRE-off periods sometimes catch genuine downturns. The 50% position bleeds while the regime is still bearish. By the time Arsenal flips back to bearish and the full veto kicks in, V6 has already eaten part of a drawdown that V5 sidestepped entirely by being flat.
I had a choice. Ship V6 instant-lift with the better headlines and accept the heavier ride. Or figure out what was actually going on and try to fix it.
The Ah-Ha Moment
I went back to those streaks I’d counted at the start.
69 days from December 2021 to March 2022. 50 days in mid-2015. 47 days late 2014. These are the long MRE-off periods. The committed bears.
But Variant A doesn’t distinguish between long bears and short ones. It fires the half-lift the FIRST time MRE goes risk-off and Arsenal happens to be at neutral. Even if MRE just flipped to risk-off three days ago. Even if it’s about to flip back to risk-on next week.
The bad half-lifts were happening in transitional periods. Chop. Whipsaw. MRE is undecided, oscillating between risk-off and risk-on, Arsenal happens to be at neutral, half-lift fires.. and gets nothing for its trouble. Maybe loses 1-2%. The position closes a few days later.
The good half-lifts were happening in the sustained bears. MRE has been firmly risk-off for weeks or months. The macro engine is committed. Then Arsenal starts to lift to neutral.. and that signal MEANS something. The bottoming process is real.
What if I gated the half-lift on duration?
Specifically: what if the half-lift only fires after MRE has been risk-off for at least N consecutive days?
I tested four gates. 15, 30, 45, and 60 days.
The 60-day gate is the answer.
It’s a Pareto improvement on both V5 AND instant-V6 on the metrics that matter most:
Mean DD when underwater: -3.16%.. better than V5’s -3.21%
Time at DD < -5%: 19.9%.. better than V5’s 20.7%
Total DD area: 161.8.. the LOWEST of any tested variant
Max DD: -13.09%.. better than V5 by 1.26pp, better than instant-V6 by 0.70pp
Sharpe: 1.027.. only 0.009 below instant-V6, +0.017 over V5
CAGR: 11.42%.. solid +0.51% over V5
The half-lift now only fires on 221 days instead of 530. The bad ones got filtered out. The good ones survived.
V6 vs V5: The Honest Comparison
V6 with the 60-day gate isn’t a transformation. V5 was already a great engine. V6 makes it incrementally better:
Sharpe: 1.01 → 1.03
Sortino: 1.10 → 1.16
Calmar: 0.76 → 0.87 (14% relative improvement.. the biggest single jump)
Max DD: -14.3% → -13.1%
CAGR: 10.9% → 11.4% (compounds to ~12% extra terminal value over 23 years)
Drawdown profile: smoother than V5, not just better at the worst case
The cost: trades go from 2.4 per year to 3.8 per year. Most of the new trades are partial entries (going from 0% to 50% during a sustained bear that’s bottoming, or back from 50% to 0% if Arsenal flips bearish again). On SPY at any meaningful size the slippage is negligible. But it’s worth knowing.
The other cost is more subtle. There will be future periods where V6 catches part of a bear market that V5 would have dodged completely. That’s the price of the recovery captures elsewhere. The aggregate trades favorably across history. But you’ll feel those moments more than you would have on V5.
I’m willing to pay that price. The +0.5% CAGR compounds. The smoother drawdown profile compounds psychologically. The Calmar improvement is real money.
The Lesson Here
The number on the spreadsheet said V6-instant was better. By the rules of how I would have evaluated this six months ago, V6-instant was a clean upgrade and I should have shipped it.
I almost did.
What stopped me was looking at the actual chart with my actual eyes.
This is the part of the work that doesn’t show up in any test suite. No forward test, no Monte Carlo, no parameter sensitivity sweep would have caught the “heavier drawdown” feeling. The metrics they check are summary metrics. They compress what they’re measuring.
They miss texture.
Texture matters. Drawdown texture is the lived experience of running a strategy.
It determines whether you can stay invested when it matters. A strategy with a -14% max drawdown that you ride through cleanly is materially different from a strategy with a -13% max drawdown and twice as much grinding mid-cycle pain.
Trust the numbers. But not blindly. The numbers know what you asked them. They don’t know what you didn’t.
Where This Lands
V6 with the 60-day gate is will be deployed and live tomorrow morning. Tomorrows 7:15AM EST run.. and every pre-market run going forward.. uses the new engine.
Please bear with me while I fix any bugs or issues as this is a large overhaul. If for any reason there are any signal changes I will confirm as always, with a Substack post.
If you’re on the dashboard, you’ll start seeing a new state in the risk panel: NEUTRAL (50%). Color-coded amber. It indicates the half-lift is firing.
Currently we’re in RISK-ON 100% with no half-lift active.. that’s normal. The half-lift is structural insurance against the next sustained bear market, not something firing today.
Forward tests passed clean across all seven. The OOS train/test split shows V6 actually OUTPERFORMS in the test period (Sharpe 0.82 train → 1.31 test).. same pattern as V5, slightly stronger.
Monte Carlo says the strategy timing has real predictive power, not lucky sequencing. Crisis stress shows V6 dodges 47% of GFC drawdown, 22% of COVID, 17% of 2022. The full battery will be in the research folder if you want to dig in.
If you check the dashboard and see a state you’ve never seen before during the next risk-off period.. that’s the half-lift telling you Arsenal sees a bottom forming while MRE is still cautious.
Half-position is the right answer in those moments. The math says so.
What’s Next
V6 will be live as of tomorrow morning.
The journey from V5 took about three weeks. The total elapsed time from “I bet asymmetric hysteresis would help” to “the 60-day gate is the answer” was maybe 12 hours of focused thinking once I started really looking.
The ah-ha moment was in the gap between the numbers and the chart. That gap is where the real work is.
“It’s only after we’ve lost everything that we’re free to do anything.”
I haven’t lost anything yet, but I almost shipped a strategy that would have cost me something.. and I would have called it an upgrade.
The eyes know. The chart knows. Trust both.
— Durden out.
✊🧼
Disclaimer: This content is for educational and informational purposes only. It does not constitute financial advice, investment advice, or a recommendation to buy or sell any asset. Trading equities and futures involves substantial risk of loss, including the potential for loss exceeding your initial investment.
Past performance, whether backtested or live, does not guarantee future results. Backtested performance has inherent limitations: it is designed with the benefit of hindsight, does not reflect actual trading, and does not account for all factors that may affect real-world execution.
The author is not a licensed financial advisor. Always do your own research and consult a qualified financial professional before making investment decisions. You are solely responsible for your own trading decisions.
Want the live dashboards behind these insights?
Subscribe on Substack Free subscribers get research updates. Paid subscribers get live macro tools + signal alerts.