How to Evaluate SPY Signal Performance

A headline win rate is not enough to evaluate a SPY options signal. Before a percentage means anything, you need to know what counted as a signal, which records entered the denominator, how long each observation lasted, and whether the result measures SPY itself or an option contract.

Astronomer's live SPY options trading signals archive publishes the underlying records so those questions can be checked. This guide explains how to read that evidence without turning an observed historical result into a promise about the next signal.

Begin with the exact question being measured

An Astronomer signal is a timestamped detection of statistically significant options activity. It identifies the underlying symbol, call or put direction, strike, expiration, detection time, and SPY price at detection. It is not a record of an order placed in a brokerage account.

That distinction determines what the archive can answer. It can show whether SPY subsequently moved in the signal's direction during a defined observation window. By itself, it cannot show the return earned by buying the flagged contract, because an option return also requires an entry premium, an exit premium, executable quotes, and trading costs.

Start an evaluation by writing the question precisely:

After each recorded signal, did the underlying SPY price move in the signal's direction between the detection minute and the recorded close in the same session?

That is narrower—and more auditable—than asking whether a signal “worked.”

Check the denominator before the percentage

On September 16, 2026, the SPY archive contained 68 signals but reported its directional result across 62 completed signals. The denominator was 62, not 68. Open observations do not yet have a recorded close. A record without the required intraday price bars cannot be evaluated on the same basis. An implausible bad-data candle should not be allowed to create a result that no market participant could have observed.

The exclusions should be stated beside the statistic and applied consistently to favorable and unfavorable records. A useful review asks:

1. How many total signals are in the archive?
2. How many have completed observation windows?
3. How many contain the required underlying-price data?
4. Which records were excluded, and under what rule?
5. Does the rule operate the same way when a signal loses?

A percentage without those counts hides more than it reveals.

Keep the observation window fixed

Time horizon can change a result completely. A call might move against its direction for three minutes, recover by noon, and finish higher at the closing bell. Each horizon answers a different question.

Astronomer's archive uses the one-minute bar at detection as the underlying reference and the bar at the observation's recorded close as the endpoint. Both occur in the same trading session. Results are not extended overnight to wait for a favorable move.

When comparing signal records, confirm that they use the same:

• entry convention;
• exit convention;
• trading-session boundary; and
• treatment of missing minutes and market closures.

Changing the horizon after seeing the price path introduces hindsight. A repeatable rule must be defined before the outcome is known.

Read directional rate with uncertainty

For Astronomer's directional measure, a completed call observation counts as moving in its direction when SPY finishes above its detection price. A put counts when SPY finishes below its detection price.

The calculation is simple:

directional rate = observations that moved in direction / completed observations

For example, 53 directional observations out of 62 completed observations produce 85.5% after rounding. That describes this sample. It does not establish an 85.5% probability that the next signal will move in direction.

Sampling uncertainty matters, especially while an archive is small. A 95% Wilson interval around 53 of 62 is approximately 74.7% to 92.2%. That interval is not a forecast or a guarantee; it demonstrates how much uncertainty remains around the observed proportion even before market conditions change.

As the archive grows, watch whether the rate remains stable rather than judging the model from one attractive snapshot.

Direction and magnitude answer different questions

A high directional rate can coexist with tiny favorable moves and occasional large adverse moves. Evaluation therefore needs both frequency and magnitude.

The archive reports an average favorable intraday peak for signals with usable bar data. This is the most favorable SPY level reached during the observation, not an executed exit. It answers, “How far did the underlying travel in the signal's direction at its best point?” It does not answer, “What return could a subscriber reliably capture?”

No rule identifies the session's peak in advance. Treat peak movement as a description of the path, not a trading result. Where possible, compare it with the move at the fixed recorded close, the median move, the range of outcomes, and the losing observations. An average alone can be dominated by a few large values.

Do not convert SPY movement directly into option profit

An option contract does not move one-for-one with SPY. Its price also reflects time remaining, implied volatility, strike and moneyness, interest rates, dividends, liquidity, and the bid-ask spread. Cboe describes delta, gamma, theta, vega, and rho as sensitivities to several of those inputs in its guide to the option Greeks.

That is why a 0.20% favorable SPY move is not a measured option return. Two contracts with the same underlying move can produce different outcomes because they have different strikes, expirations, implied volatilities, or entry prices. For a genuine option P&L calculation, you would need at minimum:

• an executable option entry quote or fill;
• a defined exit rule and executable exit quote or fill;
• contract quantity and multiplier;
• bid-ask spread and slippage; and
• commissions and fees.

Without those inputs, the defensible result is underlying movement—not option profit and loss.

Treat modeled returns as scenarios

Astronomer sometimes expresses an underlying move through a fixed leverage proxy to make scale easier to visualize. A modeled figure is a scenario, not a quoted option price, fill, account return, or promise.

The assumptions must travel with the number. A maximum modeled result is the single largest observation in the archive, not a typical outcome. It should never substitute for the distribution of all results.

Back-tested and hypothetical results can help analyze a rule, but they do not replicate live execution or predict future performance. FINRA makes the same distinction in its discussion of back-tested performance.

Segment the archive carefully

Aggregate results may conceal meaningful differences. Useful comparisons can include:

• SPY call signals versus SPY put signals;
• same-day 0DTE contracts versus later expirations;
• detections near the opening bell versus later in the session;
• high-volatility versus quieter periods; and
• recent observations versus the full archive.

Segmentation creates smaller samples, however. A subgroup with five signals can show an extreme percentage by chance. Always publish the count beside a subgroup rate, and avoid creating categories only after noticing which slice looks best.

Study the signals that moved the wrong way

An honest archive preserves unfavorable observations. They reveal failure modes that a winning percentage cannot: reversals after detection, activity that did not follow through, illiquid contracts, short observation windows, and market conditions in which the directional label carried little information.

Review losing records with the same attention given to the best result. Check the original timestamp, strike, expiration, SPY price, and price path. Do not silently redefine the horizon or remove a record because its outcome is inconvenient.

A practical evaluation checklist

Before relying on any SPY signal-performance claim, ask:

1. Is the underlying signal record public and timestamped?
2. Is the denominator shown alongside the rate?
3. Are incomplete and excluded observations explained?
4. Are entry, exit, and session boundaries fixed in advance?
5. Does the claim measure SPY movement or actual option P&L?
6. Are favorable peaks distinguished from executable exits?
7. Are averages shown with sample size and losing records?
8. Are modeled figures labeled with their assumptions?
9. Are calls, puts, expirations, and market regimes being compared with enough data?
10. Can the result be reproduced from the public methodology and archive?

The strongest evaluation is not the one with the largest percentage. It is the one whose definitions, records, exclusions, and limitations remain visible when the result is favorable and when it is not.

Read the full options signal methodology, learn how to interpret a SPY options trading signal, or inspect the current SPY options trading signals archive.

For educational purposes only. Astronomer signals are observations of market activity, not financial advice or a recommendation to buy or sell any security. Options trading involves substantial risk, including the possible loss of principal. Past performance does not guarantee future results.