I get this question a lot, Do seasonal patterns work? What about when they have less than a hundred trades? umbrella. using market regime, you can divide the market behaviour in 2-20 buckets.
The answer is yes. There are a lot of reasons why they work, and you can dig into it to find out. For example, hedge funds quarterly window dressing, rotational strategies, weather patterns, weekend risk averse, and much more. They all affect the markets in some way, and if you can find the pattern then that’s an extra edge that you can exploit to build robust strategies and add it to your portfolio.
Some seasonal patterns erode with time, new ones pop up all the time, and some patterns have been there for decades, even after many books, articles, and strategies exposes them.
Regardless of how and when you find the pattern and build a strategy around it, like all strategies, it still has to pass robustness test. In my Algo Trading Masterclass, I teach how to do robust testing on seasonal patterns. Keep in mind that seasonal patterns can be used as a strategy on their own or as a filter for another signal.
If you are interested in the US S&P 500 index then most likely you came across this pattern “Turn Around Tuesday”, the pattern suggests that buying weakness on Monday and selling strength on Tuesday usually produce good results.
Now if you have been watching my StatOasis YouTube Channel then you know it doesn't matter where or how we find out about an edge, we have to test to find out the truth. So here we go:
First let’s establish the base, below are the guidelines for our test:
Instrument: S&P500 eMini Continuous Futures Contract
Cost: No slippage & commissions are deducted
Date Range: Jan 2006 till November 2022
Entry Rules: Buy Monday on Close
Exit Rules: Sell Tuesday on Open
Other Exits: no profit target, no stoploss
It is extremely simple strategy (if we can call it that), but we get a really good picture about this edge
Don't know how to code? you can download the code for free in StatOasis Community
Base Strategy, buying end of day, selling Tuesday
1 is Monday, 2 is Tuesday .... 5 is Friday.
The results are very clear, Monday is the best day of the week by far, with more than two times better Ret/DD ratio than the second best which is Tuesday. Actually, buying Monday beats all other days in all metrics.
This is how the equity curve looks like for buying Monday and selling Tuesday
Base Strategy, buying end of day, selling Tuesday
I have longer data for the S&P 500 ETF (SPY), and the results for SPY since 1993 are extremely close to the S&P 500 Futures which shows the resilience of this pattern.
SPY ETF Base Strategy, buying Monday end of day, selling Tuesday
From here the sky is the limit, we start by adding some filters, since this is a very short-term strategy, then short term filters should work very nicely here.
One of my favorite filters is Internal Bar Strength (IBS). Check out these video on this filter:
Don't know how to code? you can download the code for free in StatOasis Community
and here are the results for adding this filter. so we are keeping base rules, but only entering if IBS filter is below 0.25 and exiting on Tuesday close
Base Strategy plus IBS filter
Another one of my favorite filters is ADX, you can check this video about the filter:
and here are the results of adding ADX greater than 20 to our base strategy
Base Strategy, plus ADX filter
Recently I introduced a mean reversion indicator that has a different take on how to calculate the pull back, it is called Casey C%, you can watch its video here:
and below the results of adding Casey C% and RSI filters respectively
Base Strategy, plus Casey C% filter
Base Strategy, plus RSI filter
Another way to look for mean reversion is to filter for lower close. In the table below you can see the different variations of going long on Monday close if it is lower than Friday close by a certain percent
Base Strategy, plus Monday close lower than Friday by a certain %
Notice how all % iterations are profitable, also notice the higher the % below Friday close the higher the average trade, higher win rate, lower drawdown, higher profit factor, but all comes at the expense of number of trades.
Like we saw above filters do work, and usually they fall under Direction & Volatility. In my Algo Trading Masterclass I provide these two filter categories under Market Regime umbrella. using market regime, you can divide the market behavior in 2-20 buckets.
Using this concept, with testing it shows that the pattern false apart in two market regimes, Quiet (Volatility) and Extreme Bull (Direction).
Base Strategy, with Market Regime Direction
Base Strategy, with Market Regime Volatility.
So we can stop trading when we encounter these markets and boost our strategy performance.
Don't know how to code? you can download the code for free in StatOasis Community
Finally, here are the results with one condition and multiple exits by surrounding our signal with 1-2 ATRs for profit target and 2-5 ATRs for stop loss and maximum number of days, so this will protect the signal up/down while still giving it a lot of time to work its magic.
Final Strategy Equity Chart
Final Strategy Performance Summary
Final Strategy Trade Analysis
Final Strategy Annual Period Analysis
Using exact same logic and applying it to longer data history trading on SPY ETF, it is nice to see that our previous decade out of sample is still showing the same behavior
Final Strategy Equity Chart for SPY ETF
Final Strategy Annual Period Analysis for SPY ETF
Final Strategy Trade Analysis for SPY ETF
Don't know how to code? you can download the code for free in StatOasis Community
Watch this video on the same strategy with different direction