ACondor Update: Trades That Open and Exits That Fill

This ACondor update is about how often the bot trades, and why it was trading less than it should have. Nothing below changes what the bot is willing to trade. The strike selection, the wing width, the volatility bar and every entry filter are exactly what they were last week. What changed is a set of rules that were stopping qualifying trades from being taken, and one that was stopping good exits from filling. The short version of every entry is on the full changelog.

The Problem: Four or Five Names Qualify, and the Bot Held One of Each

On a calm day the scan runs through about 250 symbols and three to five survive it. Those survivors are almost always iron condors on index funds. Lately that list has been the index funds, QQQ and SPY and a couple of their relatives, because single stocks have to clear a much higher volatility bar than index funds do, and in a quiet market almost none of them get there.

Then a second rule took over. The bot held one position per symbol and would not open another until that one closed. So on a day when five names qualified and the bot already held four of them, there was nothing left to do. In one two-day stretch the sandbox account passed on QQQ, SPY and SPX eighty-seven times. Every one of those was a trade that met the standard, on a name the bot had already decided it liked, refused on a technicality.

That rule was written before we knew which trades were working. The index condors have since turned out to be the steadiest part of the book, as the review of the first 122 trades showed, so the one-per-name cap was doing the most damage to the trades with the best record.

Several Positions on One Fund, Each on a Different Expiration

The bot can now hold more than one position on the same index fund, up to a limit the account sets, with one condition: each has to sit on a different expiration date. When it builds the next position it is handed the list of expirations it already holds and has to pick another one. A trade that would land on an expiration already held is refused outright.

The trade itself does not change. Same delta, same ten-point wings, same volatility requirement. A second position on QQQ is the same trade a month further out. Spreading one setup across several expirations is common practice among premium sellers because the positions come off at different times, so a single bad week does not land on the whole book at once.

Single stocks still get one position each. The bot adds at most one new position per name per scan, so it cannot pile in during a single cycle. On my live account the limit is two, on the sandbox it is five, and it is a setting like any other.

A Setting That Had Never Actually Worked

While doing this I found that the bot had been ignoring one of its own settings.

ACondor prefers monthly expirations, the third Friday, because those have the deepest markets and the tightest spreads. That preference has been switched on since July. It had never once been applied. The instruction was being passed to the part of the system that picks the expiration date, that part did not accept it, and the error was being swallowed silently. The bot fell back to picking the closest date to the middle of its target window, which frequently meant a weekly.

You can see it in the trades. A position opened September 8 went to an October 23 weekly while the October monthly sat right there in range. It is fixed, and that fallback now writes a warning to the log, so a setting cannot go dead unnoticed again.

Nothing crashed and nothing looked wrong on the dashboard, which is why it went two months without anyone catching it.

Closing Orders That Give Ground

The smallest change here is the one with the most money attached. When the bot decided to close a position, it offered the midpoint between the bid and the ask. If nobody took that price within ninety seconds the order was cancelled, and on the next pass the bot offered the same midpoint again. If the market was not going to take the midpoint, the close never happened at all.

One position on my live account shows the cost of that. It hit its take-profit rule, the closing order sat unfilled at the midpoint, and it was cancelled. By the time the bot looked again the position had drifted back below the take-profit level, so the rule stopped firing. The trade was up thirty percent of what I paid for it when the rule fired. It is now down seventeen percent, and it is still open, holding buying power that other trades needed.

Now each failed attempt gives up one tick on the next one. Pay a penny more to buy something back, accept a penny less to sell it. There is a hard ceiling tied to what the position is actually worth, so the bot cannot chase a price down. Entry orders have worked this way since August and their fill rate improved immediately. Exits should have had it then.

A closing rule that fires and then fails to fill is worse than having no rule at all, because it leaves the position on after the reason for closing it has passed, and it goes quiet instead of trying again at a price that works.

A Settings Problem That Only Shows Up on a Bigger Account

The last one matters most for anyone who is not me.

My live account is small. The sandbox account is paper money sized at a hundred thousand dollars, and it was deploying about three and a half thousand of it. Three separate limits were holding it there, and every one of them was a number that made sense on a small real account and no sense at all on a larger one: a cap on total money deployed, a cap on risk per trade that a standard index condor could not fit under, and the one-position-per-name rule above.

The caps are corrected on both accounts. The deeper issue is that those numbers were set by hand for one account and then applied to another, which does not scale to anyone else. The next piece of work is a setup that asks for account size and a risk preference, then derives the dollar limits from the account instead of asking you to know what a reasonable per-trade risk cap is. That is the right fix, and it is being designed now.

What This Does and Does Not Mean

Trading more often is worth nothing on its own. Everything above removes a rule that blocked a qualifying trade, and the bar for qualifying is where it was: the volatility floors, the liquidity checks and the strike placement are untouched. The weekly review that measures whether each change is earning its place runs against the same experiment list it did last week.

Several of these changes are days old, and a few days of results proves nothing about any of them. Every position ACondor opens carries substantial risk of loss. There is no guarantee of profit, I am not a registered investment advisor, and nothing here is investment advice.

Request Early Access and be first when we open to new accounts.