Auto Player
Players that buy and sell on their own with the user's virtual budget. Each player runs three books with the same budget: rule only, rule + AI veto, and AI only; all are compared with buying and holding the BIST 100. No real orders are placed.
Creating a player
What it is: Name, virtual budget, strategy, the stocks to play (all or selected ones), commission, and whether the AI books run.
What data feeds it: The player and player_book tables. Strategies: Evidence-backed signal · best-measured model (the reliable buy signal on Summary; if there is none, it waits in cash), L · Low volatility (the calmest stocks of the last 60 days, every 20 trading days), B · Cross-sectional ranking · 5 days (the 5-day signal of model B, every 5 trading days), W · Weighted wave · 5 days (stocks that lag behind their neighbours in the live link network and are pulled up, every 5 trading days), Market · equal weight (all stocks in equal amounts). If stocks are selected, the ranking is done only among them; at most 20 positions are held. A player makes its first decision as soon as it is created. Decisions are made only on a day whose prices are complete: while data fetching or the pipeline is running, or while even one of the companies that had a price on the previous trading day has not yet received that day's price, the player waits (at most until 09:30 the next morning, before the open, so as not to wait forever during a real trading halt; in that case the top bar keeps showing the gap); it also keeps waiting if the strategy's target belongs to a day newer than the decision day.
What it shows: The user's players; each one's strategy, budget, universe, start day, the return of the three books, and the return of XU100 over the same period.
What can be concluded from it: Players are visible only to their owner. The historical evidence for the strategies is on the Summary screen: in today's measurements no buy signal came out reliable in both periods; that is why the "Evidence-backed signal · best-measured model" player waits in cash most of the time. The other strategies run without this evidence requirement, and their results are measured live on this screen.
Three books and the comparison
What it is: Each book's value, return, difference in points from XU100 and from the rule book, cash, maximum drawdown, and a chart of daily value.
What data feeds it: The player_value table holds the closing value of each decision day; the XU100 line shows the index bought at the close of the start day and held. Decisions are made after the close, and orders are filled immediately at the (adjusted) closing price of the decision day, with commission on every buy and sell; a purchase the next morning uses yesterday's price, and whole shares are bought. Because the source gives no opening price (only the day's weighted average), filling at the next open is not possible. Since the price effect of news arriving overnight is not reflected in the purchase, the result is somewhat optimistic compared with reality.
What it shows: Rule only: the strategy itself. Rule + AI veto: the same strategy, but the AI can stop a purchase if there is clearly negative news specific to that stock. AI only: every day the AI makes its own decision among the strategy's top 40 candidates and the stocks held. The AI books wait for the day's news review or for 23:30.
What can be concluded from it: If the veto book beats the rule book, the AI's vetoes are making money; if the AI book beats both the rule book and XU100, the AI's decisions are useful. A difference over a few weeks may be chance; the comparison gains meaning over time. The cost of the AI calls is shown in the header and is paid according to your account's AI choice.
Positions, today, all trades and vetoes
What it is: The selected book's holdings, what was bought and sold on the last day, the full order history, and (in the veto book) the purchases the AI stopped.
What data feeds it: The player_position and player_order tables. Order states: waiting for price (if the decision day's price is not yet available), filled, stopped by AI, not enough cash, cancelled (if the price does not arrive within 5 trading days).
What it shows: Next to each ticker are the company name and the narrowest index it belongs to (BIST 30, BIST 50, BIST 100 or outside BIST 100); the index list is fetched from Borsa İstanbul every business day in the evening, and if it has not been fetched yet no index is shown. For positions: quantity, cost (including commission), last price, value, profit/loss and weight; for orders: decision and fill day, quantity, cash effect, commission and reason (with Claude's rationale). For vetoes: the decision-day price, the last price and the change: if the stock fell afterwards, the veto counts as correct.
What can be concluded from it: The decision log records every decision of every book and the days on which the AI could not be asked. The player checks for new data every ten minutes; for now the data are daily closes, so prices do not change during the day and decisions are made after the close.
The investment information, comments and recommendations given here are not within the scope of investment advisory services. Investment advisory services are provided under an investment advisory agreement to be signed between a client and brokerage houses, portfolio management companies, or banks that do not accept deposits. The signals here are produced from historical data with statistical models, are shown the same to everyone and are not personalised; they may not suit your financial situation or your risk and return preferences. Therefore, making investment decisions based solely on the information given here may not produce results that meet your expectations.