Orientation

**The requirements** for the state machine project are:
1. to be able to handle real-time and historical data events coming in up to one minute frequency. This relaxes some of the requirements and complexity compared to typical high frequency trading (HFT) systems.
2. Vectorized version will still be available... In quantstrat you have path-dependent (i.e. event-driven) and non-path-dependent (i.e. vectorized) option, but only path-dependent one works... So even if you have a simple strategy where a vectorised version would do the trick, quantstrat still goes through the slow, main event-driven loop... Keep and enhance vectorised version in parallel with event based state machine one.

List of external orientation **resources**, in addition to understanding current `strategery` infrastructure. Suggestion is to read the list in the order as below:

---
- [ ] http://www.quantstart.com/articles/Best-Programming-Language-for-Algorithmic-Trading-Systems
- [ ] http://www.quantstart.com/articles/Choosing-a-Platform-for-Backtesting-and-Automated-Execution
- [ ] http://www.r-programming.org/papers
- [ ] **This is the key resource**: Part I-VII of http://www.quantstart.com/articles/Event-Driven-Backtesting-with-Python-Part-I. I have full source code in Python for complete state machine model. Can share it via mail.
- [ ] `twp` package (Trading With Python, by Jev Kuznetsov) source: https://code.google.com/p/trading-with-python/source/browse/trunk/lib/
- [ ]

http://cran.r-project.org/web/packages/IBrokers/vignettes/RealTime.pdf
- [ ] `quantstrat` package - the only existing event-driven backtester in R that I know of. It is poorly implemented performance-wise and code complexity-wise I opine, but has almost all the functionality that I aim to use. They use callback model I think (not transition table).
- [ ] SIT package in R (a.k.a. Systematic Investor Toolbox) - A research-driven, vectorized backtesting framework. See Michael Kapler's blog or [github project page](https://github.com/systematicinvestor/SIT)
- [ ] other relevant R packages, namely `data.table`, `TTR`, `quantmod`, `PerformanceAnalytics`, `blotter`, `quantstrat`, `SIT` , `fInstruments` (`RMetrics`), `FinancialInstrument`
- [ ] Think of investment portfolio as a vending machine managing inventory. Our machine is bit more complicated: it buys and sells items following user-specified trading rules, maintains inventory stats and tracks resulting profit/loss). But analogy with vending machine is helpful. See: http://sourcemaking.com/design_patterns/state
- [ ] Another implementation I've seen is this: http://premium.working-money.com/wm/display.asp?art=532. I think tracking portfolio states is better approach than tracking buy/sell states as described here.
- [ ] Another inspiration: http://eng.wealthfront.com/2012/10/developing-portfolio-backtester-in-r.html

Read the listed QuantStart event-driven backtesting series, the Trading With Python source, quantstrat, SIT, and the cited R package resources in the suggested order. Then compare those approaches with the current strategery infrastructure and document the requirements for supporting real-time and historical events while retaining a vectorized path; no file or test is named.

Written by the indexing model from the issue text.