Nothing against people in India, some of the nicest people I've ever met were from India. They just happen to have an unfair competitive advantage against Americans.
You are in drawdown for 25 days and getting antsy. In your sim/backtesting testing or modeling.... What is your typical drawdown like? How long does it last on avg? What is a worst case draw down and duration in any given year? What is your sample size and how far back in years does it test?
I can't really say, my system has a significant difference in performance for t>0 than for t<0. For t<0 performance is close to perfect, for t>0 accuracy is about 60% with an average gain/loss of about 4. As far as maximum potential drawdown, it would take weeks of simulation to find out. Sample size I'm typically using 200 trading days for development, sometimes less for recent IPOs.
Sure your not from La-La land? People from India often have an advantage in getting a job because if they were educated in India they went to school either for free or almost free so they aren't burdened down with student loans. Many companies use their VISA requirements to manipulate them into working for lower wages once they are hired, ie. work cheap or lose your VISA. Not exactly a free market, but they are more attractive to employers for these two reasons.
Just as your husband left you, I am leaving your thread for you are a very sick person who is extremely nasty and bitter. Have fun losing your money to the game, I am sure it will help compound your abandonment issues.
Actually, when I come to my office, I have no friends, I am truly a ruthless son of a bitch, I don't think in nasty ways nor bitter cause I need the wannabees to increase my accounts, just like I need the person who started this journals' money. I don't ever wish upon anyone in a bad way, their uneducated ways will lose their money cause too small of sample size of one year, ROFLAMO is going to screw them bad. When I backtest methods on daily bars, I go back 30 years min which is like only 252 trading days in year times 30 years is rather small 7590 days, but seems reasonable enough. I want to test over at least four President terms to see if I see patterns to be unified in some ways. Using one year is like have dinner with half a shrimp. And if you trading stocks, one person doesn't lose and other makes money, only in futures or options have this quality. A stock could go down in value four years in a row and stock holder could have made money cause of dividends. LOL IPOs, I don't know of any bigger traders that have been trading longer than fifteen years that trade them, gunslingers trade them and often times crash and burn. I spend all my time on not losing, I don't try to make big profits, I go for very consistent small profits. On long term trading is where I go for homeruns, but even those are hedged well to not losing much. I am just an old fuddy-duddy.
Thanks for your suggestion on increasing the time span, I have tried both longer and shorter time spans than the one I am currently using. I doubt there are any similarities between our trading systems if you are back testing on 30 years of data though. I'd have to build a big cluster to get enough performance to think about going back that far but from what I have seen going back too far reduces the accuracy of my system. This has to do with the specific algorithms I am using. Trading stocks short term I am trading against mostly other short term traders, since almost none of the stocks I trade pay dividends. I'm sure there is a more formal way to quantify the change in money between short term traders than what I propose here. Here goes, to quantify the money changing hands in the short term I would integrate the absolute value of the difference between the price-volume product and a simple moving average. ie Integral([(price x volume)-SMA])dt where the SMA is over the trading duration of interest. Anyhow, I'm glad you found my system amusing. If I keep trading sideways as I have been the last 25 days, I'll have to either throw it out or fix it, perhaps I'll reduce the memory in the system and run through more data. Thanks for your suggestions.