RSI is one of the most popular technical indicators among quant traders, particularly the 2-period and 4-period RSI.
Previous analysis and articles have shown how RSI(2) maintained a strong edge in stocks through most of the 2000s.
The MFI indicator (money flow index) is similar to RSI but incorporates volume as well.
MFI is not as popular as RSI, however, MFI works just as well, and in many cases, works better than RSI for short-term trading and swing trading.
What is the Money Flow Index (MFI)?
Like RSI, Money Flow Index measures the speed of price movements. But unlike RSI, it also measures the strength of money flowing into and out of a stock. It does this using volume.
Amibroker.com has a good definition of MFI on their website which is summarized as below:
To calculate MFI you must first calculate MoneyFlow which is determined by multiplying the average price of the day by the volume.
Next, you must separate the money flows into positive and negative values to work out the MoneyRatio. If the price was up on a particular day you call it Positive Money Flow and if prices are down it’s called Negative Money Flow.
The MoneyRatio can now be normalized into an oscillator form which means it can be easily used (just like RSI) to identify oversold and overbought conditions.
When MFI is low it means selling pressure is stronger than buying pressure and vice versa. Like RSI, traders look to buy when the MFI is oversold such as below 30.
RSI or MFI – which one should you use?
To test the MFI indicator I loaded up some backtesting software with historical stock data from Norgate Data. I then ran a number of tests to compare the two indicators.
3-Day Holding Period
The following table compares RSI buy signals with MFI buy signals over a holding period of three days between 1/2008 – 1/2018 on 10,000 US Russell stocks.
No commissions are included and all entries/exits are based on the next day open price.
As you can see from the table above, the MFI indicator produced better results than the standard RSI over a holding period of three days.
The buy signal MFI(8) < 30 produced the best results with an average profit of 0.3% per trade, a win rate of 52.66% and a profit factor of 1.18 from a large sample of 232,589 trades.
This was superior to the buy signal RSI(8) < 30 which gave an average profit of 0.18% per trade and a profit factor of 1.09.
5-Day Holding Period
The following table compares RSI buy signals with MFI buy signals over a holding period of five days:
As you can see from the table above, the MFI indicator produced better results than the standard RSI over a holding period of five days.
The buy signal MFI(2) < 10 produced the best results with an average profit of 0.39% per trade, a win rate of 53.22% and a profit factor of 1.20.
This was superior to the buy signal RSI(2) < 10 which gave an average profit of 0.24% per trade and a profit factor of 1.11.
10-Day Holding Period
Since a 5-day holding period produced some clear results we can also run the same test using a 10-day holding period.
Note that a 10-day holding period usually produces better results due to the upward drift of stock prices and we don’t like to go much longer than this in our analysis:
As you can see from this table, MFI indicator once again produced the best result.
The buy signal MFI(2) < 10 gave us an average profit of 0.68% per trade with a win rate of 54% and a profit factor of 1.25.
This is the best result we have achieved so far.
RSI is a popular technical indicator among traders but MFI is rarely talked about.
However, you can see from the results that the Money Flow Index is just as good, if not better than the RSI indicator at picking oversold stocks.
These results are based on a large sample of trades across 10 years of historical data. The results also incorporate delisted stocks in order to avoid survivorship-bias.
It’s quite clear that the MFI indicator works just as well as RSI.
If you are interested in more strategies and analysis you can try our program which includes a strategy based on the MFI indicator called Money Flow 3X.
Simulations in this article produced with Amibroker using historical data from Norgate.