Open Interest Put Call Ratio: A Research Primer

Many have heard of Put/Call ratio, and many have also used it as a basis for general market sentiment. We even have it as a feature on the website, but how does this metric perform?

A high PCR signals that the market is bearish, and a low PCR signals it’s bullish. It’s pretty black and white, but if PCR exists, why don’t we see much about Open Interest PCR?

Let's take a look at this real quick.

As a note: This is by no means suggestive "alpha" or a "strategy", this is simply just a backtest on PCR vs OIPCR for curiosity's sake, and to look at a common metric used for evaluating performance.

This dataset samples 137 trading days, so a little over 6 months worth of data. I looked at SPY, AAPL, PLTR, and GS to look at some data between both OI PCR and ole’ reliable.

Why these? Well

1) Index

2) Related to index well

3) Call heavy ticker

4) Wide PCR oscillations

So here's what the charts look like with SPY's PCR and OIPCR, just for measure.

What’s interesting about the OI PCR is that it seems to drift with the market, and mirrors SPY’s movement fairly well.

Image
Image

How about AAPL?

AAPL’s OIPCR is relatively similar to SPY’s as it seems to mirror movement (and is also 6% of SPY's weight), rather than act as an oscillator (PCR).

Image
Image

But how do these perform as signals?

Well, it’s hard to set thresholds as a lot of PCR is relative to market levels, so I took the pentiles as thresholds (20/80) to gather a few more outliers.

Signal quality is calculated as a success if 5 sessions after the alert is inversely different.

NOTE: These are not returns, just black and white “if X is above/below level alerted, then success”

SPY:

OIPCR: 60% for 5 day period

PCR: 50% for 5 day

OIPCR: 80% for 10 day period

PCR: 53.85% for 10 day

Image
Image

AAPL:

OIPCR5: 47.83%

PCR5: 48.39%

OIPCR10: 50%

PCR10: 50%

Image
Image

PLTR:

OIPCR5: 45.45%

PCR5: 44.44%

OIPCR10: 40.62%

PCR10: 45%

Image
Image

GS:

OIPCR5: 44%

PCR5: 47.22%

OIPCR10: 33.33%

PCR10: 52.94%

Image
Image

As far as a signal from the 80/20 rule is concerned, there doesn’t seem to be much indicator that PCR or OIPCR are signifying anything. The charts show some well timed bottoms and tops, but there’s also a lot of misses in between.

One could dynamically adjust the PCR thresholds in real time, but unfortunately there isn’t much to these as indicators.

However, if you've found success with these indicators, let me know how you're evaluating them! I'd love to chat more.

inb4 "ackshually u could make money if you bought/sold within 1.85 days"

Yes, in a lot of these situations, you could make money on a short timeframe, but those require you perfectly timing the market, which we all know nobody can do.

So I decided I wasn’t really happy with the results from these. Yes, these are very juvenile indicators, and most likely only have a small indication on market direction, if any at all, but checking in the deciles (90th percentile/10th percent) might tell us something else.

Should these work well as indicators, the tails /might/ be more efficient, right?

Let's look.

SPY:

OIPCR5: 50%

PCR5: 83.33%

OIPCR10: 42.86%

PCR10: 80%

Image
Image

AAPL:

OIPCR5: 20%

PCR5: 40%

OIPCR10: 20%

PCR10: 33.33%

Image
Image

PLTR:

OIPCR5: 25%

PCR5: 11% (yikes lmao)

OIPCR10: 33%

PCR10: 44.44%

Image
Image

GS:

OIPCR5: 33.33%

PCR5: 50%

OIPCR10: 11.11%

PCR10: 60%

Image
Image

Okay, that's enough with these.

PCR has been long used as a general sentiment indicator. From backtests on these 4 tickers, it's not the most reliable, clearly. OIPCR seems to be even worse in the deciles (90/10).

OIPCR is clearly unreliable, and doesn't really tell us anything. If you checked more tickers, you might find a diamond in the rough where it works. You might even find it works on a longer timeframe, but as for now, this is just 6 months worth of backtesting, and it's a dud.

Regarding PCR, it's a lot more streaky. There are periods where it seems to work, and perhaps adjusting the thresholds to be tighter would yield a better signal, but that's outside the scope of this thread.

So what did this thread accomplish? Well, it was meant to be a comparison between PCR and OIPCR to check if these are meaningful in comparison to each other, or meaningful at all.

Thanks for reading! Let me know if you have any questions on Twitter https://twitter.com/falcon_fintwit