SYNAPSEA backtested edge in 0DTE SPX. One clear call all session. Trades when the edge is there.
Subscribers get two things in real time. The engine's call on SPX all session, in five plain words: PUT TREND, PUT LEAN, CHOP, CALL LEAN, CALL TREND. And the engine's own trades, from a strategy backtested on five years of SPX options, sent the moment it buys. Most days it doesn't trade, and you won't hear from it.
Backtested results are hypothetical and don't guarantee future results. Not investment advice.
Signal starts with a free trial of 7 trading days. Weekends and market holidays don't count. $0 today, and the trial keeps going until you've seen a trade. Founding price $99.50/mo for 12 months, 13 of 25 seats left, through Oct 31.
The member desk, card by card.
Example cards, built from a real session at October 7, 2026 at 13:26 ET. Live, each one recalculates through the session.
The desk
Six cards, one screen.
Members see six at once, chosen from 14 engine-powered studies. Scroll through the default six, then see how a member swaps one in.
Engine call · SPX
Called at 13:26 ETCALL
SPX is moving higher.
Strength
- PUT
- CHOP
- CALL
A new call posts once it holds for one 5-minute bar. Strength is how big and clean the move is right now, not a probability. This is what the engine called, not a recommendation. Disclaimer
01 · The call
One of five words, all session.
Call Trend, Put Trend, Call Lean, Put Lean or Chop. A new call posts once it holds for a full 5-minute bar, and the strength meter shows how big and clean the move is right now. It is not a probability.
What the engine sees
ExampleWhat keeps it at full strength: the last hour staying efficient (above 0.30) and room left to run. Session efficiency under 0.07, or a stall at an extreme the tape has already made, takes the side off the call.
Call: CALL, held a full 5-minute bar. Upside: price is moving higher. Strength 4/5.
The read is shifting off CALL. The call updates if the shift holds through 13:26.
Strength 2/5, up from 1: momentum is expanding.
Strength 1/5, down from 2: momentum is fading.
Strength 2/5, up from 1: momentum is expanding.
What adds strength: the last hour's tape turning efficient, 0.40 or better for 2 minutes, or a 0.30% thrust (about 23 pts). It falls back to CHOP if the leg gives back its move or session efficiency fades under 0.07.
Call: CALL, held a full 5-minute bar. Upside: price is moving higher. Strength 1/5.
The read is shifting off CALL. The call updates if the shift holds through 12:46.
No follow-through: the shift failed to hold. CALL stands.
The read is shifting off CALL. The call updates if the shift holds through 12:36.
What keeps it at full strength: the last hour staying efficient (above 0.30) and room left to run. Session efficiency under 0.07, or a stall at an extreme the tape has already made, takes the side off the call.
Call: CALL, held a full 5-minute bar. Upside: price is moving higher. Strength 3/5.
The read is shifting off CALL. The call updates if the shift holds through 12:30.
What adds strength: the last hour's tape turning efficient, 0.40 or better for 2 minutes, or a 0.30% thrust (about 23 pts). It falls back to CHOP if the leg gives back its move or session efficiency fades under 0.07.
Call: CALL, held a full 5-minute bar. Upside: price is moving higher. Strength 1/5.
The read is shifting off CALL. The call updates if the shift holds through 12:17.
Strength 3/5, down from 4: momentum is fading.
Strength 4/5, up from 3: momentum is expanding.
Call: CALL, held a full 5-minute bar. Upside: price is moving higher. Strength 3/5.
The read is shifting off CALL. The call updates if the shift holds through 11:55.
Strength 1/5, down from 2: momentum is fading.
The read is shifting off CALL. The call updates if the shift holds through 11:51.
Call: CALL, held a full 5-minute bar. Upside: price is moving higher. Strength 2/5.
Directional pressure building. Confirms if it holds through 11:45.
Call: CHOP, held a full 5-minute bar. No directional conviction. Price is rotating without follow-through.
Opening read. SPX down 7 pts on the session, below the prior-day low at 7,806.
02 · What the engine sees
The reason behind the word, as it happens.
The engine writes a line when something changes, a level breaks or a call starts to form, and checks in every 15 minutes when nothing does. Members read why the call is what it is, in the engine’s own words.
03 · Today’s calls
Every change, stamped.
Each time the call changed, newest first, with the minute it posted and its strength. The engine’s trade shows in the same timeline when it takes one.
Synapse Spread Finder · SPXBETA
▲ CALLFires with every CALL or PUT call: a 10-wide SPX spread, one strike in the money, held to the close.
CALL as a spread: the SPX 7,800 / 7,810 call spread at $4.80 (13:25 quote), held to the close.
Risk / spread$480Max gain$520Breakeven7,804.80Full gain above7,810Each call's spread settles at the close and builds a live record here. Call spreads have no backtest; the engine's own trade does.
04 · Synapse Spread Finder
A defined-risk spread for every call.
Every directional call is priced as a 10-wide SPX spread one strike in the money, off the live chain the minute it posts, held to the close. Each one settles at the bell and builds its own live record. Beta: call spreads have no backtest; the engine’s own trade does.
Closing imbalance · MOCBETA
Final 16:00 ETLeaning buy
Read from /ES in the minute after the 15:50 print.
/ES 15:50 → 15:51+22.50 pts/ES 15:51 → 16:00+17.00 pts/ES went with the lean into the close.
A proxy, not the imbalance itself. Synapse does not receive the exchange imbalance feed, so it reads the auction's side from how /ES moves in the minute after the 15:50 print. In 2023-26 testing, moves of 3+ points matched the real imbalance's side about 3 times in 4; smaller moves were close to a coin flip, so they show no read. It shows which way the auction leans, not where SPX goes next.
05 · Closing imbalance
Which way the closing auction leans.
At 15:50 ET the closing auctions publish their imbalance. Synapse reads the side from how /ES moves in the minute after, and only on a clear move. Beta, and labeled as a proxy: it shows the auction’s lean, not where SPX goes next. The example read is June 9, 2026, from our 2023-26 test.
Projected close · SPX
Updated 13:26 ET7,804
Range 7,788 – 7,820
Where the options market prices today’s close. A range drawn this way held the close 7 of 10 times out of sample.
The spread's breakeven (7,804.80) sits inside the range; full gain above 7,810.
The range narrows as the day runs out. It is centered on the market's own price: no direction is added.
06 · Projected close
Where the options market pins the close.
A range around today’s close that narrows into the bell, centered on the market’s own price with no direction added, with the spread’s breakeven and full-gain strikes marked on it. In out-of-sample testing a range drawn this way held the close 7 of 10 times.
07 · Make it yours
Six slots, fourteen studies.
Members pick their six from 14 engine-powered studies. Drag one from the library onto a slot and it replaces the card there; drag one slot onto another to swap them. GEX map, order flow, value area, key levels, expected move and more, and the layout saves to the member’s account.
How often you win is only half of it.
Every strategy has two numbers: how often it wins, and how big a win is next to a loss. Above the line it makes money over time. Below it, it bleeds. Drag the sliders, or tap a book to see where it sits.
WHY IT FEELS BACKWARDS
Most people judge a strategy by how often it wins, and that instinct points the wrong way here. Phoenix wins 80.3% of the time and still needs 72.1% just to break even, because its losses are bigger than its wins. Peregrine loses close to half its trades and is the stronger of the two per trade. The engine also sits out most days, so a quiet week can be the strategy working. And in the backtest Peregrine once lost 7 trades in a row without anything being broken. None of this feels right in the moment, which is why it has to be judged over hundreds of trades, not a handful.
Phoenix wins often, and small
It wins 80.3% of the time, but an average win is only 0.39x an average loss, so it needs 72.1% just to break even. The edge is the gap. Think of the casino: it wins most hands by a little and pays out big now and then.
Peregrine wins about half, and big
It wins 54.5% of the time, and an average win is 1.91x an average loss, so it breaks even at 34.4%. Think of a card counter: they wait for the deck to turn their way before they bet. That waiting is why most days Peregrine doesn't trade.
Think card counting
A good counter's edge is around one percent of what they bet, and they still lose plenty of hands. They bet more when the deck favors them and less, or nothing, when it doesn't. You can't see that in one night. It shows up over thousands of hands. One session of Synapse tells you about as much as one night at the table.
Why backtested, then walked forward
Ed Thorp worked out card counting on an IBM 704 at MIT, simulating hands before he bet a dollar, and published Beat the Dealer in 1962. A backtest is the same move: play the rules against years of real history first. The trap is grading a rule on the same history it was tuned on. It learns the noise and looks like a money printer, then fails live. Walking forward means the rules are picked on one stretch of history and graded on the next stretch they never saw, then the window rolls on. Only what holds up that way runs live.
Backtested and hypothetical: one unit of each book, Jan 3, 2022 to May 13, 2026, fills at the ask (529 Phoenix trades, 202 Peregrine). Peregrine's average win includes Apr 9, 2025, one session far bigger than any other; without it, Peregrine wins 54.2% at 1.66x and breaks even at 37.6%. Past results don't guarantee future ones. Nothing here is a recommendation. See what a bad stretch looks like, Kelly sizing and how big each trade is, and the full disclaimer.
Try Signal free for 7 trading days.
$0 today. Cancel before the trial ends and you pay nothing. Founding price $99.50/mo for 12 months, 13 of 25 seats left, through Oct 31.
Not ready for a trial? Get the call by email.
Free, no card. One email after each session's close: every call the engine made, and its trade if it made one.