Updated August 14, 2026 · Beta · DeriveAI Pro

FTO-1.5+

Better exits. Now with timing.

FTO-1.5+ starts with FTO-1's contract choice, adapts the profit target and stop, then adds a time dimension: when the trade may resolve and when an open position deserves another look.

Try DeriveAI free

The Plus update · August 14, 2026

FTO-1.5+ adds the clock.

FTO-1.5+ uses FTO-1 to choose the contract, adapts the profit target and stop, then adds something traders have always had to guess: when the trade may resolve and when an open position deserves another look.

21.7%

less timing error

The new single-option timing estimate landed closer to the day the target or stop was actually reached. It improves the estimate; it does not make a trade resolve faster.

86 in 100

landed within one day

Across resolved single-option test trades, the estimate was within one calendar day of the actual target-or-stop day 85.6% of the time.

3.9%

received an early-close review

The daily layer spoke up on a small share of trades, usually around day one. It only suggests a reassessment; it never closes a position automatically.

The new timing and daily review results apply to single options. Debit- and credit-spread policy and timing guidance remain unchanged. Timing and review figures were measured under the previous contract selector; the upgraded selector picks a different mix of trades, so they are being re-measured in forward confirmation.

Illustrative output · not a live recommendation

A trade plan you can read at a glance

The odds answer whether the target may be reached. The timeline adds when a target or stop may resolve and when an open trade may deserve a fresh look. Keeping those answers separate prevents a timing estimate from sounding like a promise of profit.

Will the profit target arrive first?

62%

estimated chance of reaching the selected profit target before the stop or expiry

This is the existing FTO-1.5 answer. Plus does not change it.

If this trade becomes a winner

When the profit may arrive

Around day 1.2
Half of similar winners had reached profit
Around day 3.4
Most similar winners had reached profit
Day 4
If it is still open, take another look
Day 7
Expiry is the hard deadline
The review on day four is not an automatic sell order.

The 1.2- and 3.4-day estimates apply only to comparable outcomes that reached the selected profit target before the stop or expiry.

What improved in testing

We gave the model only information available when the trade was selected, asked when the target or stop would be reached, and compared that estimate with what happened later. The result is a more useful clock for single-option trades.

Single options · 22,241 resolved test trades

The estimate landed closer to the resolution day.

The bars show the average distance between the estimated day and the day the target or stop was actually reached. Shorter is better.

FTO-1.5+

≈8.5 hours*

Earlier timing estimate

≈10.9 hours*

The average timing miss shrank by 21.7% — roughly 2.4 hours as an intuitive equivalent.

*This does not mean trades hit a target or stop 2.4 hours sooner. It means the estimated resolution day was closer. The source data is recorded in calendar days.

Single options · resolved test trades

About 86 out of 100 landed within one day.

That means the estimate was usually close enough to help a trader set expectations without pretending to know an exact hour.

85.6%

85.6% of resolved trades finished within one day of the estimate, up from 83.7% with the earlier timing method.

This measures timing accuracy across trades whose target or stop resolved. It is not an 85.6% win rate.

Spread guidance stays as it was.

This update changes the single-option path. Debit and credit spreads keep the policy and timing models traders were already using.

Debit spreads

Contract and exit policy

Unchanged

Timing guidance

Existing spread model

Automatic close

Never

Credit spreads

Contract and exit policy

Unchanged

Timing guidance

Existing spread model

Automatic close

Never

The updated single-option result should not be presented as an improvement to debit- or credit-spread timing.

See the exact research measurements

For technical readers: mean absolute error is the average distance between the estimated and observed resolution day. Median pinball loss is another way to grade that timing estimate. Lower is better for both.

Timing measure	FTO-1.5+	Earlier estimate
Average distance from resolution day 21.7% lower error	0.3547 days	0.4528 days
Median timing loss 21.7% lower loss	0.1774	0.2264
Resolved within one day of estimate +1.91 percentage points	85.56%	83.65%

A one-month timing test on 22,241 resolved single-option trades; 71.01% resolved on day one. The test window and its preceding month were inspected during architecture selection, so fresh forward monitoring is still required. These timing results were measured under the previous contract selector's trade mix; the upgraded selector picks a different top slice, so they are re-measured on the new selection in forward confirmation.

It predicts the resolution day better; it does not make the market move faster.
The 21.7% result means the estimate landed closer to the day the target or stop was reached in testing. It does not mean either outcome happens sooner.
Timing accuracy is not win rate.
The 85.6% figure says the estimate was within one day on resolved test trades. It does not say 85.6% of trades were profitable.
Spreads did not change in this update.
Debit and credit spreads keep their existing policy and timing models. The new timing and daily close-review results apply to single options.
A review day is not a sell order.
The entry-time model does not see the live mark, current Greeks, liquidity, remaining reward, or the latest probability path. An overdue trade should be reassessed, not closed automatically.

A real example

A SPY spread that could make $2.74 or lose $0.26. FTO-1.5 chose to hold out for 75% of the profit and bail after losing 25%. That is roughly a 31 to 1 payoff, so it only has to be right one time in twenty to come out ahead. It gave the trade a 20% chance, which sounds bad and is in fact good.

Most people would have taken profits early and set a wider stop. On that trade it would have been the worse call.

What the new exit policy changed in testing

Walk-forward validation, 12 monthly folds with a 30-day embargo: one trade per search selected by the upgraded FTO-1, exits chosen by conservative expected value. These are frictionless modeled barrier outcomes before commissions and slippage, carried by a few strong months, and pending one month of forward confirmation. They are not a profitability claim.

2×

modeled per-trade outcome vs the previous exit model

The upgraded exit model doubled the modeled average outcome per trade: +0.0215R, against +0.0107R for the previous version on the same trades.

8 of 12

months better than the previous exit model

The upgraded exit model beat the previous one in 8 of 12 walk-forward months. The result is regime-carried: a few strong months contribute most of the edge.

−0.083R

the fixed policy on the same trades

Exiting every trade at +100%/−50% averaged −0.0829R on the same selection. Choosing the exits per contract is where the improvement comes from.

Read the full evidence report

What it learned from

FTO-1.5 does not learn from quotes alone. It learns from real optimizer recommendations graded to resolution, each one replayed under nine different take-profit and stop-loss plans to see which exits would have served it best.

11.6M

training rows behind the exit policy

11,648,898 rows: 1,294,322 graded contract recommendations, each scored under nine take-profit and stop-loss plans.

1.29M

graded recommendations behind those rows

Every recommendation carries a resolved outcome: target reached, stopped out, or expired. Earlier history is excluded rather than graded with information the model could not honestly have had at the time.

exit plans scored per contract

Take profit early, late, or in between; cut losses tight, loose, or in the middle. The plan with the best conservative expected value wins.

It understands spreads.
Debit and credit spreads get their own dedicated models rather than being squeezed through one built for single contracts. It reads the whole position and both legs.
It cannot fool itself.
The model is built so a harder target can never be given better odds than an easier one, and a wider safety net can never be given worse odds than a tighter one. Those would be nonsense, so it is not allowed to say them.
It tells you the bar, not just the number.
Every probability now shows the win rate that trade needs just to break even. A 55% chance sounds strong until you see it needs 60%.
It says when nothing is worth trading.
If none of the nine plans makes money on any candidate, the screen says so rather than serving you the least bad one dressed up as a pick.

Contract choice still comes from FTO-1.
FTO-1.5+ uses the upgraded FTO-1 ranking (0.737 walk-forward ROC-AUC, up from 0.688) to choose the contract, then applies its exit policy, timing estimate and daily review guidance.
It is for picking, not for trading everything.
Positive modeled averages in testing are thin, frictionless and carried by a few strong months. Commissions, slippage and sizing are not included, so the model remains a way to pick the strongest setups, not a reason to trade every search.
This update does not improve spread models.
Debit- and credit-spread policy and timing models are unchanged. The new timing and day-1+ close review apply to single options.
It is in beta.
It is good enough to use and not finished. We would rather say that than pretend otherwise.

Choosing the exits is where the edge shows

Across the 12-month walk-forward test, the upgraded exit policy turned the modeled all-trades average positive while the fixed exit plan lost on the same selection. Frictionless modeled outcomes before costs, regime-carried, and not a profitability claim. The strongest setups also stay protected: the edge guard that keeps exit experiments off the very top of the ranking was rebased to the upgraded score.

All selected trades · adaptive policy

+0.0215R modeled

Same trades · fixed +100%/−50% policy

−0.0829R modeled

Spreads, before and after it chooses the exits

Average return as a share of what you put at risk. Above the line is a profit.

Fixed

Chosen

Debit spreads

Credit spreads

The numbers, in full

Walk-forward validation across 12 monthly folds with a 30-day embargo. R is measured against entry premium. These are frictionless modeled barrier outcomes before commissions, slippage, liquidity and execution effects: not live or guaranteed returns, and not a profitability claim. The positive average is regime-carried and has not yet been confirmed on a fresh forward month.

Measure	Previous exit model	Upgraded exit model
Average modeled outcome per trade One trade per search, selected by the upgraded FTO-1, exits chosen by conservative expected value.	+0.0107R	+0.0215R
Months better, head to head Which version produced the better modeled average, month by walk-forward month.	4 of 12	8 of 12
Policy-row ranking accuracy (AUC) How well each version orders individual exit plans. The upgraded model gives up a little ordering precision for better trade outcomes.	0.7610	0.7555
Fixed-policy reference The same selected trades exited at +100%/−50%, for comparison against both versions.	−0.0829R	−0.0829R

Spreads, average outcome per trade

Strategy	Fixed exits	Chosen exits
Debit spreads	−0.0128R	+0.0378R
Credit spreads	−0.0287R	−0.0067R

Getting it

FTO-1.5+ comes with DeriveAI Pro. Pick it from the model menu in the Optimizer, or use search, where Pro accounts get it automatically. Single options receive the updated timing and daily review guidance; debit and credit spreads keep their existing policy and timing models. Spread optimizations remain available to everyone, including free accounts.

See FTO-1

Try DeriveAI free