Fractional (fixed) order size <1 does not work

[nikolhaas65]

Hi,
I want to backtest with order size of fixed amount (fixed_bet). Previous suggestion was to use fixed_bet/self.equity or fixed_bet/self.data.Close[-1]. Such sizes produce no trades and zero pnl.
I tried self.sell(size=0.05) or self.buy(size=0.5), but result is the same - zero.
After deep dive I think the problem is in class Broker where the size_float is converted into the size_int inside _process_orders.

size = order.size
if -1 < size < 1:
    size = copysign(int((self.margin_available * self._leverage * abs(size))
                        // adjusted_price), size)
    # Not enough cash/margin even for a single unit
    if not size:
        self.orders.remove(order)
        continue
assert size == round(size)
need_size = int(size)

I am trading PERPETUALS, where BTC index price is around 100_000 USD today, but I can trade them at notional as small as 10 USD. I set initial cash at 1_000_000, so I have enough equity. Nothing helps. And the reason, I think, is above.

PS. After trades I am forced to re-scale all trades, which is inconvenient, because I have to rebuild statistics.
PPS. Yes, I have seen earlier issues and replies, but do not find then satisfactory.
For example, this solution produces zero order_size.

current_price = self.data.Close[-1]
order_size = 100 // current_price

Thanks.

[kernc]

Previous suggestion

Yes, I have seen earlier issues and replies

Can you point them out?

Issue #226 (comment) is about wanting to order assets for a fixed amount of cash ($100).