Commit Graph

247 Commits

Author SHA1 Message Date
fawce 1f04ee4ece committed a set_trace... 2012-04-23 21:04:51 -04:00
fawce fe793b503d hotpatching a merge error (duplication of the max dd check). 2012-04-23 20:50:08 -04:00
fawce 54d3579ceb added boolean to results for exceeding max losses in a single simulated day. 2012-04-23 15:10:44 -04:00
fawce 1d9953255a added stop out for max drawdown. 2012-04-20 23:41:23 -04:00
fawce bab0e2bd19 made distinction between cumulative behavior and daily behavior a bit more clear. 2012-04-20 15:08:39 -04:00
fawce bc14e7e3b7 zipline, now a cold, heartless, http://open.spotify.com/track/1xshgoh575otNXRfeYgh9D
Pretty fast too...
2012-04-20 12:21:03 -04:00
fawce 28c380c245 removed heartbeat on every iteration, resulted in a 50% reduction in processing time per day.
However, the big lesson from this experiment is that the major bottleneck is in the feedback loop. By eliminating the feedback loop and instead putting the transaction simulation into the client end of the zipline, I think we could accelerate by 10x.
2012-04-20 12:21:03 -04:00
fawce 6f5e9b0559 fixed documentation 2012-04-19 23:32:37 -04:00
fawce 08a5bd5a0d re-arranged fields so that history includes daily snapshots for cumulative measures. 2012-04-19 23:27:35 -04:00
fawce ddb1c51569 moved epoch_now to date_utils 2012-04-18 23:14:21 -04:00
fawce 68040bad9a consolidated some of the date util methods. added iso8061 support 2012-04-18 17:11:01 -04:00
fawce 1b5b92d75f dropped unwanted fields. 2012-04-18 12:36:46 -04:00
fawce 6561292e4c some efficiency changes - no longer transferring the list of daily returns, no longer holding transactions in the daily and the cumulative performance periods. 2012-04-18 12:36:46 -04:00
fawce 7eb0ba67ac clarifying the results protocol for the backtest. 2012-04-18 12:36:46 -04:00
fawce 309f78a030 modification of the transport protocol -- keeping it close to the export of data from the performance tracker. 2012-04-18 12:36:46 -04:00
fawce 5fd30216e5 revised protocol to maintain original structure. 2012-04-18 12:36:46 -04:00
fawce 2fcb68f59e Merge branch 'master' of github.com:quantopian/zipline 2012-04-18 12:36:32 -04:00
Stephen Diehl 864899003b Clean up test case. 2012-04-18 12:29:40 -04:00
Stephen Diehl 28f975fd55 Don't replace right now. 2012-04-18 11:19:10 -04:00
Stephen Diehl b3de0a2fcc Added ndict test suite. 2012-04-18 11:18:48 -04:00
Stephen Diehl e2638f1cbb Fixed arguments on magic methods. 2012-04-18 11:01:12 -04:00
Stephen Diehl daaa45e9c1 Namedicts edges fixed, and awesomer ndict 2012-04-18 10:53:04 -04:00
fawce 9db18846c9 fixed type-os in serialization code 2012-04-16 13:37:54 -04:00
fawce bd616a9ec0 eliminated special handling for non-blocking components. 2012-04-16 12:06:39 -04:00
fawce 02d7f0a4c8 heavy work on the feedback loop from trade client to order source. tests are passing using a busy wait inside the trade client. hopefully we can find a more elegant approach. 2012-04-14 15:09:52 -04:00
fawce 4486dd0cad removed some straggling test code. 2012-04-13 16:54:49 -04:00
fawce 38982fcdab major fix is with the non-blocking behavior of order source. also fixed time-compression in the trading client. 2012-04-13 15:08:17 -04:00
fawce dfc3523197 fixed default to be partial fill 2012-04-13 09:58:12 -04:00
fawce 1559a84c7b added simulation style to transation simulator, to facilitate tests. Fixed roll-over bug in max cap and max leverage calculation. 2012-04-13 09:53:20 -04:00
fawce aea2e1189c fixes to the calculation of transactions and associated tests for long and short orders. 2012-04-12 10:46:10 -04:00
fawce b78097241a rounded out tests to cover more trading cases (spreading, collapsing, expiring) 2012-04-11 12:00:16 -04:00
fawce f759cac61b added a simple test for the transaction simulator, heavily revised the simulator as a result. 2012-04-10 22:46:50 -04:00
fawce db7a4d2f91 Merge pull request #29 from quantopian/fawce_sprint1
added transactions to the daily update objects.
2012-04-10 11:17:37 -07:00
fawce 368341ce61 moved transaction store to PerformancePeriod. added the position data to the performance message. 2012-04-10 14:12:03 -04:00
fawce 8ce5159e91 Merge pull request #30 from quantopian/stable_upstream
Stable upstream
2012-04-10 08:00:39 -07:00
Stephen Diehl ea340b6186 Datetime subclasses with tzinfo=UTC by default 2012-04-10 08:30:37 -04:00
Stephen Diehl 6f27009e82 Started datetime utiles lib. 2012-04-10 00:30:53 -04:00
fawce 2eeb92442e added transactions to the daily update objects. 2012-04-10 00:02:54 -04:00
Stephen Diehl 04d389b774 Remove old pandas-zmq protocol. 2012-04-09 23:43:53 -04:00
fawce 73d63057c1 Updated ordersource to be concurrency agnostic! 2012-04-09 11:10:41 -04:00
fawce 14166ccc30 dropped the extra days in trading range... 2012-04-09 10:31:11 -04:00
fawce 57c39bf615 switching to only calculate the returns and risk on market close, rather than per trade. 2012-04-09 10:20:08 -04:00
Stephen Diehl 1dbe0975fc NumpyChannel 2012-04-09 07:55:11 -04:00
fawce aa2bcdf83e ending values for portfolio, equity, and cash in cumulative were removed, because they are redundant to the same values in daily. Also removed starting cash, as it is unchanging and equal to the capital base. 2012-04-08 21:09:42 -04:00
fawce 237b42c11e switched reporting to provide cash, equity, and total portfolio value 2012-04-08 13:53:40 -04:00
Stephen Diehl 16ceb68c7c Move zmq topology mapper to dev/ folder 2012-04-07 09:58:50 -04:00
fawce 37a8bda4b2 fixed bogus initial portfolio value (should be zero). 2012-04-06 22:31:35 -04:00
fawce 30dfc86ba9 fixed dates front to back to be proper market open/close, and to use start/end first_open/last_close from the TradingEnvironment. 2012-04-06 20:49:56 -04:00
Stephen Diehl 80bfbd5dcb Performance reports take either socket address or socket. 2012-04-06 13:31:39 -04:00
Stephen Diehl 5aee03212d Replayable error log. 2012-04-06 12:39:27 -04:00