Carrying a Tested Plan Into a Bot: The 1:45 Sandwich Automation

Automation is not where a trade plan begins; it is where a tested plan becomes repeatable. The video opens with the most common misunderstanding in automated trading: expecting the bot to discover the edge, make the judgments, and fix discipline. The answer given is blunt — the bot is the discipline layer that comes after the plan, not a shortcut around building one.

The lesson sits inside a three-step framework, and this is step three: automate the plan. First the strategy is defined, then it is tested in a backtester, and only then is it automated. Reverse that order and the bot is just guessing with better software.

The running example is the so-called 1:45 sandwich, chosen because it already has structure. It has a fixed timing idea, a defined strategy, and explicit entry and exit rules. Because it was tested first, nobody has to guess what the bot should do — the job is already specified.

The analogy used is a coffee maker set up the night before. The machine never decides what coffee you want; you choose the blend, add the water, set the time and strength, and in the morning the machine simply executes the setup. Bots work the same way: they do not replace the trader, they run the structure the trader built.

The practical advantage shown is creating the bot directly from the backtest. Symbol, timing, position type, entry and exit rules, and position criteria travel with it, so the scanner and monitor logic does not have to be rebuilt from memory. The trader's question changes from how to rebuild everything to whether the bot still matches the tested plan.

That review question is the heart of the lesson: automation must carry the tested plan forward, not silently become a different strategy. If the backtest traded one position at a time at a fixed hour, the bot must not scan all day or open five at once. Every boundary — capital allocation, concurrent positions, daily limits — has to be set before the bot runs.

A vague bot is treated as dangerous because it can do too much; a bot with a clear one-sentence job is easy to audit. The example job is explicit: follow the tested 1:45 p.m. sandwich strategy with the same entry, position, and exit rules from the backtest. Simple, specific, reviewable.

Before anything goes live, the bot runs in paper trading. Since the bot executes exactly what it is told, the paper phase exposes wrong criteria rather than bad luck. The approach is deliberate: automation may be built to run hands-off, but it earns live capital only after a review-and-refine pass.

The entry scanner is walked through decision by decision: has a position already been opened today, is it Monday, Tuesday, Thursday, or Friday, does the market close today since this is a same-day expiry setup, is the VIX filter between zero and twenty satisfied, and is price above the ten-day exponential average. Only a clean pass through every gate opens anything.

The position itself is a same-day SPX iron condor in a single contract: the short put five dollars under the underlying, the long put five further down, the short call ten above the short put, and the long call five above that. Pricing walks up to four quotes with a few cents of slippage tolerance, the trade is held to expiration, and an extra filter requires reward-to-risk above one hundred percent. A trial run plus the log trail then confirms the 1:45 scan firing and each check passing, and the viewer homework is to convert one favored backtest into a paper bot and reach the point of explaining every open and close from memory.