Some code to let you know the state of a partially filled order. cls_opn_crs()
oos = get_open_orders() for s in oos: for o in oos[s]: # loop thru open orders if cls_opn_crs(context, o) in [1, 3]: # if opening, not closing if o.filled < o.amount * .10: # if partial fill and very small cancel_order(o.id) # cancel remaining
The idea is that an order so thinly traded on open (at some threshold you determine) can also have a difficult time closing which can be very bad if the price is moving in the wrong direction when the time comes. You can improve the example by keeping track of the number of partial fills, then doing the math and have it operate on their mean.
About that 1 and 3 in the list [1, 3] above in checking the return from cls_opn_crs():
1: A straightup opening order, from 0 shares, it is a simple increasing number of shares, or for example an increase of allocation from 100 shares to 300 shares, or -50 shares to -112 shares.
3: Crossover opening portion. For example where a position was 100 shares and an order was entered for -500 shares (targeting -400), and at least 100 (or -100 depending on how you think of it) of the previous position have been filled, now passing 0 and into negative territory. Was long, is now short and increasing number of short shares (more negative). This is now an order that is opening (greater magnitude in the number of shares).
0: Closing order, reduction in shares toward 0
2: Closing toward zero however merely the closing portion of a crossover order
A change from -50 shares to -35 shares (or 50 to 35) is considered a "closing" order (0) here even though the target is not 0.
def cls_opn_crs(c, o): # c = context o = order object # Whether order is closing, opening or crossover (short to long or reverse) if c.portfolio.positions[o.sid].amount * o.amount < 0: # close or crossover if abs(c.portfolio.positions[o.sid].amount) < abs(o.amount - o.filled): if abs(c.portfolio.positions[o.sid].amount) - abs(o.filled) < 0: return 3 # crossed 0 shares and now opening else: return 2 # cross closing else: return 0 # closing else: return 1 # opening
Why am I sharing something so valuable?
Live trading ended in Sept 2017 apparently related to a 3% fund downturn and the number of funded algorithms after 30 months of contests were perhaps only around 15. Not to be too dramatic but I used to be worried about the competitive environment here and now feel it may be more of a question of overall platform success. Plenty of headroom for any quality algorithms to be adopted. I don't need to be holding back so much on helping you (presumably my competition, meh, maybe not really). Plus there's always a chance that your feedback can help me too.
So I plan to be posting some (not all) tips and tools I find useful. This is one of the best, I use it a lot. Actually I have it return 0 if o.stop or o.limit because my stop and limit orders are only closing orders.