How to quantify a put spread?

Why do you need to quantify a put spread

The problem with subjective trading is that the conditions for execution are different each time. Today you think the valuation is high so you make a trade, tomorrow you chase in because you read a piece of news, the day after you let a real opportunity slip because you're too busy to calculate the odds.

Quantitative trading is about fixing the decision rules, letting the machine measure all opportunities with the same ruler every day.

Actually, the core of quantitative trading can be summed up in one sentence: trading is a science. As long as you believe in science, believe that the market can be (probabilistically) predicted, and then study it using scientific, empirical methods, you understand quant trading. Whether you're doing high-frequency or low-frequency trading, whether you use statistical models or AI, whether you use an automated order system or manually list your house for sale doesn't matter. — Li Xinye's "Guide to Dating Married Women"

A good bear put spread must simultaneously satisfy three dimensions: valuation, relative strength, and odds.

1. Valuation Deviation: The price must be significantly higher than your fair value

You can't possibly assign an absolutely precise fair value to a stock, but you can set a lower bound you're willing to bet it will revert to.
For example, using my own valuation model, I determined that $55 is the walking price for INTC. Then $82.5 is abnormal.

Don't pursue a precise valuation model; pursue a clearly quantifiable state of "expensive."
A simple division (price/fair price) exceeding 1.45 means entering the firing range. This is more suitable for options trading than any DCF.

2. Relative Sector Overheating: Exclude the beta rise of the entire sector

Still using INTC as an example, INTC rising 20% doesn't necessarily mean it should be shorted. If SOXX rose 25%, it's actually underperforming.
Conversely, INTC going crazy alone while the sector stays still—this kind of divergence is most valuable. Because what you want to trade is the stock's alpha bubble, not the semiconductor industry's beta.
A very simple strategy: use the standard deviation of the 20-day return difference relative to the historical mean. This essentially asks: Is the current degree of divergence among the top 2.5% extreme cases in the past two years? Z > 2 is this type of signal.

This is a minimalist strategy. You can layer on top of it, further processing this Z-score factor, enhancing it in three directions: first, convert it to a historical percentile to eliminate outlier effects and retain extreme degree information; second, introduce divergence duration as a decay weight, e.g., after several consecutive days of Z>1.5, breaking through 2.0 multiplies signal strength; third, cross it with IV skew, only adopting the signal when the implied volatility surface shows left skew.

3. Odds Enforcement: Max Profit / Max Loss ≥ 2.5

Many people only look at direction, not odds.
The advantage of a put spread is precisely to lose small and win big. If the cost rises too much, the advantage disappears. You must write the odds threshold into your system.
For example, 2.5x is the bottom line. Below that, the system directly discards it, not even entering it as a candidate.

Moreover, calculate cost using conservative prices, not the mid-price. Because retail investors looking at option prices like to look at the mid-price, i.e., mid = (bid+ask)/2. But in reality, you buy at the ask and sell at the bid. The mid is just an illusion.

For a put spread:

· Cost to buy the long put = ask
· Income from selling the short put = bid
· Your true net debit = long_ask − short_bid

Finally, always use the worst executable price as the calculation basis.
If, based on this conservative price, the odds are still ≥ 2.5, then after actual execution, your odds will only be better, not worse.

4. Only select contracts with good liquidity, don't touch garbage

Options are not stocks. If an option has no volume, no open interest, it's easy to get in but hard to get out.

You can write a very simple filter rule. You can adjust it later yourself, for example:

· Bid-ask spread ≤ 15% of mid-price
· OI > 500
· Daily volume > 50 (preferably > 100)

Liquidity is the lifeline of stop-loss. When the market crashes and you want to close your position for profit, if no one is buying that put, you can only watch your paper profits evaporate.
It's better to give up a high-odds but illiquid contract than to be trapped inside.

5. Manual confirmation prioritized, automated execution conditional

This is the most critical decision point. Longbridge API does not support atomic submission of combo orders. If you want full automation, you can only execute legs separately.
But separate leg execution has a fatal risk: if the order is wrong, you'll be naked shorting a put.

If you insist on automation, you must enforce this order:

· Opening: First buy the long put, wait for confirmation, then sell the short put.
· Closing: First buy back the short put, then sell the long put.

Never sell the short put first.
 

Additionally, automated execution only runs during the following periods:

· No trading for 30 minutes after market open (wait for spreads to tighten)
· No new positions opened within 20 minutes before market close (avoid overnight risk)
· Only reduce positions, no new openings, 3 days before earnings

6. Position sizing based on loss amount, not nominal principal

Retail investors often say, "I'll use 5% of my capital for options." The problem is, the loss on options is not the nominal principal, but the premium.

For a put spread, maximum loss = net debit × 100 × number of contracts.

Your risk limit is "how many dollars you can lose at most," not "how much nominal principal to use."
The system first calculates the max loss for a single spread, then divides "total capital × 3%" by the single spread loss to get the contract count. This way, even if you're wrong a few times in a row, it won't hurt you badly.

7. Closing is not waiting for max profit, but exiting in stages

The time value decay of options is not linear; it decays fastest in the last two weeks. More importantly, when the price actually falls to the target, your put spread has already made most of its profit, the remaining profit space is small, and gamma risk actually increases.

So you need to actively take profits, not greedily chase tail-end profits.
Set several price targets and close in batches:

There's also a logic-based stop-loss: If the reason for your initial short (e.g., poor fundamentals) is disproven, such as a major customer order appearing or new capacity coming online, then unconditionally close the position.
This is not a technical stop-loss; it's a logic-based stop-loss.

8. Every trade must go into the database

An unrecorded trade is a trade that never happened.
SQLite can meet the needs. The fields don't need to be many; the key is being able to answer these questions:

· Which strike combination (75/55 vs 75/60) has a higher win rate?
· Do entries with odds greater than 3 actually perform better than those with odds of 2.5?
· Does entering one week before earnings worsen performance?
Without historical data, you'll never know which of your rules is an illusion.

As long as you can implement these rules in any language, any tool (Excel works, Python works, Python is of course recommended more since vive coding is so developed, it's easy to put together a set), you're already more professional than 90% of retail investors.