How do you expect anyone to answer your questions? You provide next to no information. Derive the source of your alpha and investigate why it's not there during the period in question. For more we need more information.
If I weren't so lazy I would quote you a nice Greek phrase, so just take it in English: You can not step into the same river twice. In other words, times are a changin'... So if you had a bullish strategy and back tested it 1990- 2000, geee, surprise, surprise, it worked!!! Then you are baffled why suddenly it doesn't work in the next 3 years??? Because certain strategies only work under certain market conditions, and when those change, there goes your money (and the idea that you are a great trader)... Don't feel so bad, lots of people fall into this fallacy, when they have a good strategy for the correct time. That is not a problem itself, not realizing that is. In short, your strategy isn't robust. Edit: Fuck it, here it goes: δὶς ἐς τὸν αὐτὸν ποταμὸν οὐκ ἂν ἐμβαίης.
Very incredible how things changed in 2007/08. The question is : would it be better to architect a strategy using a 20y time series or a 10y time series..... I don't think is an obvious question as is true that the strategy would be more robust, but is also true that probably would take in account price behaviour that don t exist anymore
%% Good wise points;1997 to 2007, looks like a big part of the problem, not enough bears in that trend to really weight that data .If i was worried[I'm not worried , myself]about 1997-07 again , i would find something else i could do in a period ,like that . Good thing for me , not many markets will not chop slop sideways[aka barbed wire range ]for 10 years. NOT a prediction.
You should build the strategy for your own intended time frame. Are you a scalper or an intra day trader or a swinger? Do you plan to hold stocks for months? The price fluctuates a lot, specially nowadays, thus one can be bullish on a 5 mins time frame but bearish on the hourly and still bullish on the daily and all 3 can be right at the same time! Depending on the holding time, all 3 views can make money. But markets do change too, and some bullish strategy that worked 10 years ago might not work so well today. Or it has to be adjusted, like wider stops because of volatility,etc.
Did you just keep adding indicators and clicking optimize till you got your equity curve? My feed back is in the title of the image. "Curve fit monster from hell"