What the Circuit Breaker Arrives Too Late to Fix
Your circuit breaker halts new trades at minus 10 percent drawdown. Correct behavior. Now look at the state you are in when it fires.
Four open positions averaging 2.3 percent, 9.2 percent gross exposure. Some winning, some losing. You cannot open new hedges. You cannot take offsetting positions. You are locked down with an unbalanced book. The winners close into their targets. The losers stay. You drift to minus 12. The breaker holds, but you are sitting in losing trades you can no longer manage, because the system slammed into a hard stop instead of throttling down on the approach.
Circuit breakers are reactive by nature. They catch you after you are already in trouble. The better question is why you were running at full activity while the account was deteriorating in the first place. The breaker fired because you were overtrading relative to capacity. That mismatch is the real failure, and it has a preventive fix.
Two Kinds of Capacity
Every trading system lives inside two capacity constraints at once.
Technical capacity is how hard you are trading right now. Gross exposure: the sum of open position sizes over available capital. Thirty thousand in capital against forty-five thousand in open notional is 1.5x utilization. Pure function of the current book.
Substrate capacity is how much the account can currently absorb. A function of present drawdown, daily friction (fees plus estimated slippage), and margin state. An account at minus 8 percent paying real daily fees has less substrate capacity than a flat account with the same headline capital. The ledger number is not the whole story.
When technical capacity outruns substrate capacity, the account is overextended. Positions added in that state are not feeding a healthy system; they are loading weight onto a structure already under stress. That is where cascades are born: losses compound against high exposure, open positions move against you, and the system reaches its hard breaker without ever having reduced risk on the way there.
The Draconis pattern enforces correspondence between these two measures before the breaker ever has to fire.
The Pattern
Compute both measures continuously:
```python
P_Tech = sum(abs(pos.size) for pos in open_positions) / capital
current_dd = (current_equity - peak_equity) / peak_equity
P_Bio = 1.0 - abs(current_dd) - (daily_fees / capital) - (daily_slippage_estimate / capital)
```
Then apply two rules before any new sizing:
Hard veto. If P_Tech exceeds 0.8 and P_Bio drops below 0.7, set the position multiplier to zero. No new entries. This catches the dangerous mismatch directly: above 80 percent utilization while the account has absorbed enough friction and drawdown to leave only 70 percent substrate. That gap does not hold.
Soft throttle. If P_Tech exceeds 0.6 and P_Bio drops below 0.8, set the multiplier to 0.5. New positions size at half allocation. This is the early-warning state, where overextension is approaching. Cut now, before you need the veto.
At normal conditions the multiplier is 1.0. Existing positions are never force-liquidated by this pattern; they run to their own stops and targets. Only new entries are throttled.
Why This Is Not Just a Circuit Breaker
A circuit breaker asks one binary question: is drawdown worse than the threshold? Yes, halt. No, continue. No middle state.
The Draconis pattern asks a relational one: is my current activity proportionate to my current capacity? It measures the gap between what you are doing and what the account can sustain, and it applies graduated reduction continuously rather than binary halting at a single trigger.
Watch the difference in a deteriorating market. Under a breaker alone, you keep entering full-size at minus 3, minus 5, minus 8, then stop dead at minus 10. You added exposure the whole way down and then froze.
Under Draconis, you begin entering at half size the moment P_Bio falls below 0.8. Fewer positions open through the bad stretch. When the market steadies and P_Bio recovers, the multiplier resets to 1.0 on its own. If conditions keep worsening to the veto threshold, you stop entirely, but you arrive there carrying less open exposure than you otherwise would.
The circuit breaker is the last line of defense. Draconis is the graduated response that lets the last line of defense stay unused.
A Worked Scenario
Monday, normal. Signals fire. Four positions at 2.5 percent each, 10 percent total. P_Tech 1.0, P_Bio 0.95. No rule triggers. Everything sizes at 100 percent.
Tuesday morning. An overnight deleveraging event drops several assets. Positions underwater, drawdown minus 4. P_Bio falls to about 0.91. P_Tech still 1.0 because nothing closed. The system is watching, not yet throttling.
Tuesday evening. Another leg down, drawdown minus 9. P_Bio about 0.86, P_Tech creeping up as fees accumulate. Now P_Tech is past 0.6 and P_Bio is nearing 0.8. The next proposed entry trips the soft throttle. It sizes at 50 percent.
Wednesday morning. Market steadies, drawdown recovers to minus 5. P_Bio climbs back above 0.8. Multiplier resets to 1.0.
Without the throttle, Tuesday evening would have meant full-size entries into a system already running hot during an active drawdown. The throttle does not erase the loss. It keeps you from stacking weight onto a stressed structure during the exact window when that mistake is most expensive.
Calibrate the Thresholds to Your Book
The values in the pattern, 0.8 over 0.7 for the veto and 0.6 over 0.8 for the throttle, are working values chosen for a specific strategy's position count, fee structure, and drawdown frequency. The principle is what transfers: measure the ratio of activity to capacity, and reduce proportionally when it runs too high.
Calibrate against your own loss history. Pull your worst drawdowns and ask where P_Tech and P_Bio diverged, when the throttle would have fired, and whether it would have helped. Set your thresholds from the answer.
Log every throttle event: P_Tech, P_Bio, the multiplier applied, and the proposal that triggered it. Decision logging and governance are the same practice.
The Principle
The system that survives is not the one that trades hardest in good conditions. It is the one that cuts activity proportionally as conditions worsen, so that when conditions are worst, exposure is lowest. That is not conservatism. It is the only rational posture.
Circuit breakers are necessary. But a system that leans on its breaker to save it has already failed at every earlier point where it could have throttled gracefully. Build the graduated response. Let the breaker be the net, not the first responder.
Architecture beats scale. Correspondence between activity and capacity beats raw signal throughput.