Данная статья рассчитана на программистов, проявившим интерес к использованию SQL в своих проектах. В статье читателям представляется функциональность SQLite, а также рассматриваются ее преимущества. Статья не требует знание функций SQLite, но минимальные знания SQL приветствуются.
I am pleased that running the ticks tester (from now on I will always refer to real ticks) GBPJPY, GBPJPY, GBPUSD and USDJPY data has been downloaded. This means that the arbitrage tester galleys can be thrown out of the Market!
It is too early to make any statements yet, we still need to work out how to easily bring the historical symbol bases on the brokers' servers to a near-perfect state.
How far in history will the real ticks be for testing (for your server and for the brokers' servers)? From what date is it possible to start testing? How to determine this for your broker/clearing house?
Compared multicurrency testing on opening prices and on all real ticks from January 1 to March 18. Previously optimised by opening prices for the year.
on opening prices
by real ticks
Calculation time
1 minute and 43 seconds
33 minutes
Total trades
6 055
5 102
Tiki
9 098 913
15 745 121
Relative drawdown on balance
25%
27%
Relative drawdown on funds
45%
38%
Net profit:
29 678
15 783
In general, the strategy still works, but the profit halved. The balance curve shape is about the same, but somehow it jerks a bit more on real ticks. What was missing inside a minute now works.
The halving of profit is quite significant, so we need to do a new optimization.
But the increase in calculation time is catastrophic. Apparently 1 annual pass will take 2 - 3 hours. And optimizing on my computer, for 10000+ passes - 2 - 3 weeks (at opening prices it took 12 hours).
Apparently the cloud will load 100%... Just have to wait until all agents have updated terminals.
By the way - all agents in the tester are still visible. Maybe we should only show those which have updated and can test on real ticks.
Enclosure:
Test on the real in six months gave a profit of 8 times less than at opening prices.
I'm an API. And we'll attach the base without problems.
Why wait? It's been around for a while.
https://www.mql5.com/ru/articles/862
and what do hedge and netting accounts mean, please explain?
Netting accounts are (until yesterday) MetaTrader 5 - it was possible to have a position on an instrument in one direction only - either BUY or SELL.
Hedged accounts are MetaTrader 4 - it is possible to have a position for a symbol in both directions - either BUY or SELL.
Now (as of yesterday) MetaTrader 4 can be retired as being unnecessary.
It is too early to make any statements yet, we still need to work out how to easily bring the historical symbol bases on the brokers' servers to a near-perfect state.
How far in history will the real ticks be for testing (for your server and for the brokers' servers)? From what date is it possible to start testing? How to determine this for your broker/clearing house?
Now (as of yesterday) MetaTrader 4 can be retired, as it is no longer needed.
Hedging is MetaTrader 4 - you can have a position on an instrument in both BUY and SELL directions at once.
Now (as of yesterday) MetaTrader 4 can be retired, as it is no longer needed.
Compared multicurrency testing on opening prices and on all real ticks from January 1 to March 18. Previously optimised by opening prices for the year.
In general, the strategy still works, but the profit halved. The balance curve shape is about the same, but somehow it jerks a bit more on real ticks. What was missing inside a minute now works.
The halving of profit is quite significant, so we need to do a new optimization.
But the increase in calculation time is catastrophic. Apparently 1 annual pass will take 2 - 3 hours. And optimizing on my computer, for 10000+ passes - 2 - 3 weeks (at opening prices it took 12 hours).
Apparently the cloud will load 100%... Just have to wait until all agents have updated terminals.
By the way - all agents in the tester are still visible. Maybe we should only show those which have updated and can test on real ticks.
Enclosure:
Test on the real in six months gave a profit of 8 times less than at opening prices.