Thread: ESP Timer's Account Talk

Turtle Trader

I used the Turtle Trader rules with the TSP Trading restrictions. It avoided major drops but only resulted in a 206% gain over twenty years. It missed some obvious runs which leaves me suspicious of my code which came from several old projects that I hacked together. I am going to spend some time cleaning it up to see if I have any errors. It might be interesting to add a net worth graph so people can easily grasp how volatile a strategy is if you are looking at your account value.

P.S. Open the image in a new tab to see the entire twenty years.

turtle.jpg

I calculated Buy and Hold gains for all funds that existed on 5/31/2003.

Hold.PNG

Small Cap stocks did the best so I will work on strategies for that fund. I added Win Rate, Max Loss and Trades. They don't matter for Buy and Hold but will help in choosing better strategies in the future. I am changing the Backtest engine to use the G Fund for cash so it still grows when not in another fund. Once that is done I will compare some basic Moving Average and Turtle Trader strategies. I expect them to do poorly but the journey of a thousand miles begins with a simple SMA cross.

Monday is the best day for TSP Bond funds and the worst day for TSP Stock funds. Tuesday is the best day for Stock funds. So, yes, Turnaround Tuesday is a small but real thing.

stats.PNG

I calculated the average percent change for each trading day of the month. All stock funds tend to drop at the beginning of the month and rise thirteen trading days into the month (OPEX?).

Day Of Month Stats.PNG

I calculated the average change per trading day of the year for the Stock funds and the Fixed Income fund. They look random with a positive bias. I didn't see anything useful even after adding a 10-Day Moving Average.

Stocks Change Per Day Of Year.PNG

Fixed Income Change Per Day Of Year.PNG

I now switch between the G-fund and the Target fund to show more realistic returns. Here are stats for Buy and Hold G-Fund, Buy and Hold S-Fund and the best seasonal period for the S-Fund since 2003. Total Return is the percentage gain, Win Rate is the percentage of trades that were profitable, Max Loss is the largest losing trade and Trades is the number of trades executed.

Seasonal Stats.PNG

The best seasonal strategy was one that avoided the Covid meltdown. That was a unique event that won't happen again (or won't occur in the same time period) so I need to figure out a way to screen out non-repeatable strategies. Anyway, here is the signal chart for this strategy since it is an interesting way to see the trades (open it in a new tab to view the entire twenty years).

Seasonal Signals.jpg

I finally had some time to code so I made my charts similar to StockCharts.com (multiple symbols per graph, multiple graphs per chart, variable sized graphs and opaque symbols). I am still looking for an easy-to-read background color. Anyway, now I have what I need to test Sy Harding's Seasonality System from the 90's.

stockcharts.jpg

Maybe I should save a lot of time and let ChatGPT do the trading.

Sy Harding's Seasonal System

The International Fund was the only fund that beat Buy & Hold using Harding's system and it only did that because the 2008 crash occurred during the cash period.

stats.PNG

Here is a chart showing the trades, new features and a vanilla color scheme that seems easier to read (to me).

harding.jpg

Turtle Trading from the 80's couldn't beat Buy & Hold and Sy Harding's Seasonal System from the 90's failed as well. Avoiding once-a-decade crashes appears to be key to beating Buy & Hold.

Added Percent from SMA graph which shows how far the price is above/below an SMA (in percent). Might be useful for timing or overbought/oversold indicator.

percent 50.jpg

percent 100.jpg

percent 200.jpg

I switched to a table/chart format with the goal of analyzing dozens of symbols in a few minutes instead of the hours that it currently takes me. My program downloads 30 years of stock data, sorts everything by Weekly RSI, loads it into a table that can be sorted on any column and displays the seasonal trend along with as much history as I can fit on the screen. Change is the last trading day’s change. RSI is daily RSI and RSIW is weekly RSI. SMA % is the percent from the 200 SMA (rubber band). Red cells indicate oversold and green cells indicate overbought so symbols with multiple red cells are buy candidates (and vice versa).

tsp.png

I found ETF's that use the same indices as the TSP funds: G Fund = SHV, F Fund = BND, C Fund = VOO, S Fund = VXF and I Fund = VEA. G/SHV is a problem as bond etf's distribute their income creating a saw tooth pattern. I don't have access to mutual fund data so I guess I will have to live with it until I can find a short-term government stock/ETF that retains income.

shv.PNG

I will eventually merge my old cycle program into the new program to help predict future cycle lows.

cycles.png

Are there any other features that I should consider to make it better/different than Stockcharts.com?

P.S. Open large images in a new tab to see the entire thing.