Edges
0DTE Strike Selection: The Base-Rate Test Your Contract Must Pass
Most 0DTE strike selection happens backwards. A trader forms a directional view, looks at the chain, and picks the strike whose premium feels affordable. The question never asked is the only one that is pure arithmetic: how far does the underlying have to move for this specific contract to be worth anything, and how often does this specific symbol actually move that far?
Both halves are knowable before you click. The first is a subtraction. The second is a base rate you can look up. Together they set a hard ceiling on your win probability — a ceiling that assumes you also get direction, timing and the spread right.
The two-step test every 0DTE strike must pass
Step 1 — compute the move the contract needs. For a put, break-even at expiry is strike − premium, so the move required is:
For a call it mirrors: ((K + P) − S) ÷ S. Nothing about volatility, delta or your read enters here. It is subtraction.
Step 2 — look up how often that symbol delivers that move. Not "can it" — how often it has, measured across 260 daily sessions. That number is your ceiling.
Run your own contract through both steps:
Reachability: how much move a session actually gives you
The table below is 260 daily sessions per symbol. open→low is the downside a put can reach intraday; the ≥X% columns are the unconditional base rate of a move that size happening at all.
| symbol | median day range | median open→low | ≥0.5% | ≥1% | ≥1.5% | ≥2% | ≥3% |
|---|---|---|---|---|---|---|---|
| MSTR | 5.21% | 2.63% | 95% | 84% | 72% | 61% | 43% |
| COIN | 5.00% | 2.02% | 92% | 77% | 63% | 51% | 35% |
| MU | 4.85% | 1.91% | 83% | 73% | 61% | 48% | 35% |
| AMD | 4.16% | 1.72% | 86% | 70% | 57% | 45% | 27% |
| PLTR | 3.96% | 1.67% | 85% | 70% | 54% | 43% | 28% |
| TSLA | 3.41% | 1.39% | 83% | 61% | 46% | 36% | 19% |
| NVDA | 2.71% | 1.25% | 81% | 59% | 45% | 30% | 11% |
| META | 2.31% | 1.12% | 79% | 56% | 37% | 22% | 8% |
| GOOGL | 2.30% | 0.95% | 70% | 48% | 31% | 19% | 6% |
| AAPL | 1.89% | 0.77% | 63% | 39% | 20% | 10% | 3% |
| MSFT | 1.87% | 0.82% | 68% | 40% | 27% | 13% | 7% |
| QQQ | 1.23% | 0.55% | 53% | 27% | 13% | 6% | 2% |
| SPY | 0.85% | 0.38% | 40% | 14% | 4% | 2% | 0% |
Read one row against another and the same trade stops being the same trade. A contract needing a 1% move is a coin flip on TSLA (61%) and close to a lottery ticket on SPY (14%). Identical thesis, identical structure, 4.4× different odds — decided entirely by which ticker is printed on the contract.
This is the arithmetic behind the liquidity-first view in our comparison of the most liquid 0DTE scalping tickers: tight spreads matter, but a symbol that does not move cannot pay a contract regardless of how tight its market is.
A worked failure: the read was right and the trade still lost everything
On 31 August 2026 a trader bought a TSLA 357 put at $0.60 with spot at 364.65. The thesis: price ran up all morning, expect a pullback.
The pullback came. Price ran 365.81 at 10:36 down to 361.85 at 10:55 — a real, clean 1.08% move, well within a normal TSLA session. And the contract expired worthless, because it needed a move roughly three times larger than the one the thesis correctly predicted.
That is not a directional failure. The read was fine. The instrument required an abnormal session to cash a normal read, and the base rate said so before entry: 36%, and that 36% is measured from the open, while this entry was mid-morning near the highs of an up-day, making the true conditional probability materially worse.
Why not just diagnose it as "I traded counter-trend"?
Because that diagnosis was tested and did not hold up. Trend alignment — "don't fight the day's direction" — was measured across 38,897 signals and 2,703 sessions using five separate causal direction proxies at three timeframes. The pre-declared contrast came back at +0.027 R with a Holm-corrected p of 0.34: not supported. And +0.027 R is smaller than the round-trip cost of expressing it. The counter-trend story feels explanatory and produces no action. Reachability produces an action before the click.
Strike distance: the pop is real, and so is the floor
Here is where the standard advice — stay near the money — needs refining rather than repeating. Price the same TSLA ladder at the implied volatility solved from that trader's own real $0.60 fill, then value every strike at the pullback low that actually occurred:
| strike | premium | delta | break-even move | value at 361.85 | P/L on exit |
|---|---|---|---|---|---|
| 357 | $0.60 | −0.15 | −2.26% | $1.13 | +89% |
| 360 | $1.23 | −0.26 | −1.61% | $2.12 | +72% |
| 362.5 | $2.06 | −0.38 | −1.15% | $3.30 | +60% |
| 365 | $3.20 | −0.51 | −0.78% | $4.81 | +50% |
| 367.5 | $4.66 | −0.64 | −0.50% | $6.62 | +42% |
At the moment the move happened, the far strike had the largest percentage gain — +89% against +50% for the near-money contract. That is convexity working exactly as advertised, and it is why "always trade near the money" is too blunt a rule to be true.
The catch is what happens next. The far strike has no floor: hold it and it goes to zero. The near strikes retain intrinsic value at the same spot. Move the sliders and watch the two columns diverge:
- **Committed to exiting fast** — minutes, on the move — a cheaper out-of-the-money strike maximises the pop. But the exit is not optional, it *is* the thesis. There is no floor to fall back on.
- **Any chance you will hold** — the strike must pass the reachability test, or holding *is* the loss.
The TSLA 357P failed as a mismatch, not as a strike: a scalp instrument held like a position.
Unconditionally — averaged over all outcomes rather than conditioned on a move that already happened — expected value still falls monotonically with strike distance. The far-out lottery version of this trade was tested at 0 of 80 expiries and 0 of 485 contracts. The pop is real; you finance it with frequent total losses. Conditioning matters, and conditioning on a move you have not had yet is not available to you at entry.
Why higher delta is cheaper than it looks
Strike distance is not one decision. It is three, and they move together — which is also why moneyness changes your effective leverage so sharply:
- Reachability — distance raises the move required, and the base rates above collapse as it rises.
- Delta — a $0.03 bid-ask spread on a 0.15-delta contract is a far larger cost in risk-unit terms than the same $0.03 on a 0.60-delta contract, because you divide by delta to express it as underlying movement you must win back.
- Gamma — the convexity you are paying for is concentrated near the money, and the relative spread is tightest there.
Paying up for delta buys reachability and cuts your cost-in-R at the same time. The cheap contract is usually the expensive one.
The pre-trade gate, in order
Reachability is the first of seven checks, and it is first because it is free to compute and disqualifies more trades than anything else on the list.
If any single check fails there is no trade — not a smaller trade, no trade. Sizing down a contract that cannot reach its break-even does not fix the arithmetic, it just loses less slowly.
The takeaway
Before your next 0DTE entry, do the subtraction and the lookup. It takes fifteen seconds.
- Compute
(spot − (strike − premium)) ÷ spotfor a put, or((strike + premium) − spot) ÷ spotfor a call. - Look up that percentage in your symbol's row. Below roughly a 50% base rate, pass.
- Compare the required move against the symbol's median open→low. Above 1×, you are asking for a bigger-than-typical session — viable only with a written fast exit.