Hello there:
I have a trading Idea… and I also have an idea to back test it…
As you probably know… Finviz is a popular free stock screener… where you can find stocks that fit certain criteria… I was thinking about the following strategy:
- A stock above SMA(20) ( simple movime average of 20 days).
- A stock that has at least 2 higher lows, close above prior close; close above opening price (yes it is trend following, or momentum strategy… ).
- RSI indicator less than 70
- Dollar Volume above 500 K dollars on the day (for liquidity reasons) (what I mean here is multiply the volume on the day… by the closing price… and take that as dollar volume…).
All the above can be spotted with finviz… now that I have spotted those results… I was thinking about using python to download OHLCV daily data from the last 5 years… ( using yfinance for example ) and as you may also probably know… use pandas dataframe to check all the days that the stocks fullfilled all the criteria above, and when they are…use a column called "Signal" to check this… ( that means; when ALL the conditions above are satisfied… the "Signal" column should be 1 and 0 otherwise…) .
Once the Signal is equal to 1; I was thinking about buying the next day at the open… and sell it the next day also at the open… So to test this… whenever Signal is equal to 1; calculate the ratio between the opening of the next days (for example if Signal was on Monday; calculate the returns of the days when there ws a signal … that is the ratio between the wednesday's open and Tuesday's open… ) in order to test if the strategy has worked in the past…
Once the returns are calculated multiply them all and check the results… also check how many losses there were ( return less than 1), and how many wins there were (return greater than 1) …
those stocks that have the eplaned returns greater than a certain number…( let's say 2)… and they have the number of losses greter or equal to the number of losses… consider that the strategy was successfull in the past… and trade it ( using the same rules described above…)
What do you think about this ?? can this work ?? maybe I need to modify something… maybe you can give me any suggestions about this…??
Thanks…
Hi Ghery,
In theory, it sounds like a great idea, but obviously only a backtest can tell you objectively if the strategy performs well or not. Also, after backtesting, you can try paper trading it for sometime before deploying the strategy in the live markets.
Looking forward to others' reply on your ideas as well.
Thanks.
Hello there:
Thanks for your comments on my post, now Do you think 5 years of data are enough to try this…?? maybe I need more time… maybe 10 years?? …I have already done all of the above… and found some interesting results… ticker DXCM ( NASDAQ: Dexcom inc) has a coumpounded return (what I mean with that is the product of the individual returns calculated above…) greater than 2… and also has a win rate of more than 50% …as I stated above… I calculate the number of returns greater than 1, (wins) and the number of returns less than 1 (losses)… what I found was 164 wins and 136 losses… ( in the results I replaced 0,94 everytime the return fell below that… to simulate a stop loss 6% below from where I bought it)…
and Yesterday (01/11/2023) … this stock fitted exactly all the criteria above… Do you think this might be good to try?? (as Stated above… get in at the opening… to do that place a market buy order right before the market opens with a stop loss 6% below … and place a market sell order tomorrow right before the market opens… of course if the stop loss has not been hit before…)
Can this work ??
Hi Ghery,
You can try backtesting it for 10 years as you would understand how the strategy performs in various market regimes. You can also break this 10 year data in two parts of 5 years and see whether the performance is similar in both parts.
Remember, move forward only when you are extremely confident of your trading strategy.