DETERMINISTIC BLACKJACK DECISION ENGINE RESEARCH
PHASE 3.100 — SELECTIVE-GATE RELEASED-SESSION DECOMPOSITION
LAYMAN RESULTS EXPLANATION

WHAT ARE “RELEASED SESSIONS”?

These are the sessions where:

the ordinary £160/25% trailing rule WOULD have stopped,

but

the Phase 3.99 £140 severity gate said:
“the trigger bankroll is still at least £140 — allow this session to continue.”

Phase 3.100 examines only those sessions.

No new policy is being tested.
No threshold is changed.


THE RESULT IS MUCH STRONGER THAN I EXPECTED

PRIMARY

111 sessions were released.
That is only 2.22% of all 5,000 sessions.

Of those:

90.99% subsequently reached £200.
95.50% finished with MORE money than unconditional trailing would have retained.
Only 0.90% finished with less.
None finished below £100.
None exhausted to £0.


HOLDOUT

100 sessions were released.
2.00% of all sessions.

Of those:

88.00% reached £200.
93.00% finished with more money.
Only 2.00% finished with less.
None finished below £100.
None exhausted.


THIS EXPLAINS PHASE 3.99

The selective gate changed only a very small population:

2.22% Primary;
2.00% Holdout.

But those were unusually strong continuation candidates.

That is why overall £200 attainment rose from:

14.48% to 16.50% Primary;

13.90% to 15.66% Holdout.

The gate was not indiscriminately restoring play.

It released a small, highly productive subset.


AVERAGE BANKROLL IN THE RELEASED GROUP

PRIMARY

Unconditional trailing would have retained:
£155.81.

Selective continuation actually finished:
£206.76.

Average gain from release:
+£50.95.


HOLDOUT

Unconditional:
£154.36.

Selective:
£207.31.

Average gain:
+£52.95.

The result replicates extremely strongly.


WHY IS THE AVERAGE ABOVE £200?

The £200 target is checked after a completed blackjack round.

A winning round can carry the bankroll beyond exactly £200 before termination.

So an average final of £206–£207 among this highly successful subgroup is legitimate.


WHAT DID CONTINUATION COST?

It reintroduced approximately:

£79.87 additional wagering Primary;
£80.00 Holdout

per released session.

But the additional maximum drawdown was surprisingly small:

£1.35 Primary;
£1.20 Holdout.

So for this particular released population, the extra exposure was productive far more often than destructive.


AN IMPORTANT RESULT: REFERENCE AND SELECTIVE FINALS MATCH

Average final bankroll in the released population:

PRIMARY
Reference £206.76
Selective £206.76.

HOLDOUT
Reference £207.31
Selective £207.31.

That occurs because once the selective gate releases these sessions, it does not introduce another stop later in this architecture.

Their subsequent path therefore follows the unchanged reference trajectory.

This is also an excellent deterministic implementation check.


DOES THIS PROVE £140 IS OPTIMAL?

NO.

It proves something narrower and scientifically stronger:

The £140 gate, selected from the replicated Phase 3.98 separation and frozen before Phase 3.99, successfully identified a small subgroup for which continuation was overwhelmingly beneficial in these deterministic Primary and Holdout cohorts.

We still must not search £139, £141, £145, etc. and call the best retrospective number optimal.


WHAT HAVE PHASES 3.94–3.100 ESTABLISHED?

The branch now forms a coherent chain:

3.94
Hard exits protect bankroll but destroy most £200 opportunity.

3.95
A trailing/high-water signal can preserve continued opportunity.

3.96
£160/25% trailing becomes a genuine middle-ground intervention.

3.97
Paired analysis reveals good saves versus premature exits.

3.98
Observable trigger severity separates those futures.

3.99
A predeclared £140 severity gate prospectively restores £200 opportunity.

3.100
The sessions released by that gate are shown directly to be overwhelmingly productive continuations.


IS THIS A GOOD POINT FOR A PLATE?

YES.

This is the first point where the entire branch can be explained cleanly on one visual plate.

The plate should NOT imply profitability.

Its central message should be:

HARD STOP
strong protection / severe opportunity loss

TRAILING STOP
middle ground

SELECTIVE £140 GATE
releases only stronger trigger states

RELEASED GROUP
~88–91% reach £200
~93–96% finish higher than unconditional trailing
0% £0 in both cohorts

while retaining the warning:

population return remains negative;
no guaranteed profitability;
no card counting or hidden information;
£140 is validated as a predeclared gate, not proven optimal.


WHAT SHOULD FOLLOW AFTER THIS PLATE?

I would treat 3.100 as a natural MILESTONE for this branch.

The next experiment should not immediately tune the gate.

A sensible next branch could examine robustness/generalisation of the selective architecture under a genuinely new deterministic cohort/seed set, or move to another remaining brainstorm area.

That decision can be made fresh after the plate.


PHASE 3.100 VERDICT

RELEASED POPULATION:
Small — 2.22% Primary / 2.00% Holdout.

REACHED £200:
90.99% / 88.00%.

FINISHED HIGHER THAN UNCONDITIONAL TRAILING:
95.50% / 93.00%.

FINISHED LOWER:
0.90% / 2.00%.

£0:
0% / 0%.

AVERAGE GAIN FROM RELEASE:
+£50.95 / +£52.95.

EXTRA WAGER:
~£80.

EXTRA MAX DRAWDOWN:
only ~£1.2–£1.35.

CONCLUSION:
The predeclared severity gate successfully released a small, strongly productive continuation population.

MILESTONE:
YES — selective trailing architecture now has observational, prospective causal, and paired decomposition evidence.

PLATE:
YES — recommended now.
