Lab Notes: 88capital – Rules-Based Automated Investment Strategy with AI
Some links in this article are affiliate links to products I actually use. If you sign up through them, I may earn a small commission — at no extra cost to you.
This is not financial advice — it reflects my own approach and research.
A basic rule of investing is “buy low, sell high” – but, unless you’re able to track market swings continuously, it can be difficult to recognise the right moments to buy or sell. It’s even harder to ignore those instinctive, ‘gut-feel’ decisions – when volatility hits, psychology pushes you to do the wrong thing, making this an often expensive approach to managing a personal portfolio.
I built 88888888capital, shortened to 88capital for convenience ;), in response to that problem – an automated investment strategy with AI, as a personal portfolio experiment that replaces reactive investing with an automated rules-based system using n8n, Claude, and Slack.
I wanted a hands-off system that supports systematic investments when markets dip. It is not a high-frequency trading bot, nor a get-rich-quick scheme, but an investment framework designed to run in the background – a strategy that also doubles as a content and digital product experiment for Triple 8 Labs.
What 88capital Actually Is
The core of 88capital is a signal-driven dip-buying framework designed to complement a personal portfolio, realised as an n8n automation stack. Instead of guessing, I use defined entry and exit logic based on technical indicators. If the system does not generate a signal, then no action is taken – there are no discretionary decisions. The automation layer that coordinates all this is based on workflows using Google Sheets, Yahoo Finance, Anthropic Claude (can be replaced by OpenAI, Gemini or Mistral), and Slack.
Beyond the framework, 88capital is also a content experiment. The entire methodology – from broker selection to the specific logic for the signals – is documented and published openly, and the full set of 12 “ready to import” workflows is available for purchase from Gumroad.
The Automation in Practice
The monitoring and reporting for this investment strategy with AI is handled by four scheduled n8n workflows, each of which is responsible for a distinct job: scanning for entry signals on two different watchlists, tracking open positions, and synthesising everything into a weekly report.
Workflow 1: Daily Event Scan (Mon–Fri, 08:30)
Each weekday morning the daily scan checks a curated “event watchlist” — a focused list of tickers where a near-term catalyst or elevated volatility makes a dip more likely to be meaningful. For each ticker, the workflow fetches the current price and applies a Tier 0 gate: if the price is 15% or more below the 52-week high, the ticker qualifies for full technical evaluation. Only tickers clearing this threshold continue to the next stage, fetching the 14-day Relative Strength Index (RSI) and 200-day Simple Moving Average (SMA) and running the signal evaluation logic.
Signal Assessment via Claude
Signals that pass the Tier 0 gate and technical evaluation are routed through the Anthropic API to Claude, using my n8n LLM routing workflow which allows switching between different AI models. A structured prompt containing the merged price and technical metrics goes in; a concise summary explaining what the signal means and whether the setup fits the entry criteria comes out. Only TIER 0 CANDIDATE and CAUTION alerts are posted to Slack — the daily scan also posts a notification summary regardless of whether any signals fired.
At the moment, no trades are executed automatically. When a qualified signal lands in Slack, background checks need to be run to confirm there are no recent profit warnings, no analyst consensus shift to Sell, and that there are no active governance issues. The system decides when something is worth looking at; but as recommended for all AI generated output, you need to review and decide whether to act.
Workflow 2: Weekly Watchlist Scan (Monday, 07:00)
Where the daily scan watches a focused event list, the weekly scan runs across the “main watchlist” — a broader universe of large-caps monitored on a slower cadence. It runs every Monday morning before the market opens. For every active ticker it fetches price, full technical indicators, and trailing dividend yield history, then evaluates signals using the same logic as the daily scan. Noteworthy results are posted to Slack along with a scan summary. This schedule reduces noise while ensuring nothing on the main list drifts unnoticed for longer than a week.
Workflow 3: Positions Tracker (Mon–Fri, 09:15)
The positions tracker runs each weekday shortly after market open and updates the current price of open positions in Google Sheets (using a filter since some fund types are priced less frequently). This is what keeps position data current; without it, the weekly digest would be reporting on stale prices. It also sends Slack alerts for DIP_BUY positions that hit exit or watch thresholds.
Workflow 4: Weekly Digest (Sunday, 08:00)
A digest workflow runs every Sunday morning at 08:00. It reads positions, signals, and watchlists from the portfolio data in Google Sheets and calls Claude to produce a structured Slack report covering five sections: portfolio snapshot, signal activity for the week, an opportunity radar of tickers closest to triggering, a strategy pulse, and a look-ahead at relevant market events.
Why Build An Automated Investment Strategy with AI
The primary goal is to support a strategy for building positions as part of a personal investment portfolio that requires minimal active management. By offloading monitoring to n8n and contextual analysis to an AI model like Claude, the decision when an alert arrives is reduced to a binary choice: run the pre-defined checks, then either place a trade or do nothing. Less room for emotion to influence the call.
The secondary goal is to use this experiment for content and as a digital product. The monetisation anchors are a packaged n8n workflow bundle sold via Gumroad and affiliate relationships with tools in the stack. Affiliate economics work at scale, so content volume matters.
To be honest about where things stand: building the automation was the enjoyable part. Even with AI-assisted drafting to reduce the friction, writing, editing, and publishing the documentation is the ongoing, less glamorous work. As of July 2026, the investment strategy with AI and technical automation are fully operational – the public content phase is just beginning.
The full technical details, documentation, and workflow bundle are available on the 88capital experiment page.
