# Double Simple Moving Average Crossover - Backtesting SP500

Lets look at the S&P500 and try will the backtesting method again. Will the double simple moving average crossover method generate better results as a trend follower then the other methods? We will again compare the yield of the Buy&Hold strategy with our trend follower. Again we are using data from the last 15 years (3750 trading days) and timewindows from 1 to 1000 days for the long and short simple moving average. Figure 1 shows the 3D plot and figure 2 the topview.

Compared to the DAX backtest, the distribution of the yield performance is similar: very short simple moving average pairs are not good, like long pairs. The golden triangle lies like in the DAX backtest in the combination of middle SMAs. The best combination reaches 17.44 % p.a and is the pair 228 / 204. This not as high as the DAX result, but the topology of the S&P500 yields has fewer turbulences. If you invest into an middle range SMA pair in this backtest, the danger of reaching a local minimun is not as high as with the DAX, because the mountains have a bigger surface und not so sharp edges. This shows again that a trend following method may strongly surpass the results of a simple buy & hold tactic.

### Comparison with Buy&Hold on trading the S&P500

Lets remove all yields below 6.5% p.a. - this is the buy and hold performance in this benchmark.

In fig. 3 you see that the simple moving averages which beat the benchmark lie in a triangular shape and inside this there is a ‘golden’ triangle in which the maximum results are located. The whole is similar to the DAX performance, aber the regions are more homogenous.

Figure 4 and 5 show an zoom into the golden triangle where the top SMAs lie. Fig. 6 shows an performance comparison between the best simple moving average and the S&P500 index.

### Optimal simple moving averages against yield underperformance on trading the S&P500

Lets look at the negative parts of the moving average trading system, meaning the underperformance as negative distance from the index. Figure 9 shows the underperformance for all timewindows, fig. 10 shows again a zoom on the golden triangle. You see that in the triangle especially the short pairs have a strong negative part. In comparison to the DAX backtest we see there are less underperformance parts in the golden triangle.