AI for investing: great with data, useless at predictions
AI gets data right and predictions wrong: how to use it for investing without losing money (or your mind).
English · Original discussion in Spanish · Published
AI as a stock advisor: useful for data, useless for predictions
A machine forecasting the exact movement of a stock is fiction; a machine speeding up stock research is real. The key is knowing what to ask of it: data and context, never the definitive answer. The experiences of investors with decades in the trade agree on one point: AI gets it right when asked for concrete figures and fails when asked about the future. One cited case: a tool recommended waiting until June because interest rates would rise, and they didn't.
What AI does well: gathering and organizing
Used as a data search engine, AI performs well. It can give you last year's dividend, the P/E ratio, relevant news about a company, or a quick introduction to an unfamiliar sector – such as rare earths – without reading ten reports. To get the most out of it, it helps to specify the sources: Simply Wall St., Tikr, roic.ai, the 10K filings from IR. With that, it saves hours of reading and lets you get up to estimulante ilegal in minutes.
What it doesn't do well: predicting
AI has no opinion, only statistics. It tends to agree with whoever is asking and can pick up fake news. The most seasoned investors are clear about it: nobody can say whether a stock is better at 10 than at 10.5 euros. But the tool also makes forecasting errors, like the interest rate one. And in the digital ecosystem, supposed gurus promising to multiply investments – from 3,300 to 33,900 euros in months – proliferate to harvest contacts on social media; that isn't artificial intelligence, it's reeling in the gullible.
The method: narrow down, verify and be skeptical
Some suggest using it as a mass data processor with defined criteria: low P/E, PEG below 1, dividend yield above the sector average, or debt reduction in recent years. But any automation requires a checker: an agent that monitors the responses, because AI's non-deterministic behavior can ruin an apparently routine process. The mistake would be to trust it as if it were an oracle. With all due proportion, nobody would eat what their toaster recommends.
The question that separates the saver from the speculator remains: is AI an extension of analysis or an excuse not to think? The data suggests that, without supervision, it only multiplies the estimulante ilegal at which you can lose money.
A machine forecasting the exact movement of a stock is fiction; a machine speeding up stock research is real. The key is knowing what to ask of it: data and context, never the definitive answer. The experiences of investors with decades in the trade agree on one point: AI gets it right when asked for concrete figures and fails when asked about the future. One cited case: a tool recommended waiting until June because interest rates would rise, and they didn't.
What AI does well: gathering and organizing
Used as a data search engine, AI performs well. It can give you last year's dividend, the P/E ratio, relevant news about a company, or a quick introduction to an unfamiliar sector – such as rare earths – without reading ten reports. To get the most out of it, it helps to specify the sources: Simply Wall St., Tikr, roic.ai, the 10K filings from IR. With that, it saves hours of reading and lets you get up to estimulante ilegal in minutes.
What it doesn't do well: predicting
AI has no opinion, only statistics. It tends to agree with whoever is asking and can pick up fake news. The most seasoned investors are clear about it: nobody can say whether a stock is better at 10 than at 10.5 euros. But the tool also makes forecasting errors, like the interest rate one. And in the digital ecosystem, supposed gurus promising to multiply investments – from 3,300 to 33,900 euros in months – proliferate to harvest contacts on social media; that isn't artificial intelligence, it's reeling in the gullible.
The method: narrow down, verify and be skeptical
Some suggest using it as a mass data processor with defined criteria: low P/E, PEG below 1, dividend yield above the sector average, or debt reduction in recent years. But any automation requires a checker: an agent that monitors the responses, because AI's non-deterministic behavior can ruin an apparently routine process. The mistake would be to trust it as if it were an oracle. With all due proportion, nobody would eat what their toaster recommends.
The question that separates the saver from the speculator remains: is AI an extension of analysis or an excuse not to think? The data suggests that, without supervision, it only multiplies the estimulante ilegal at which you can lose money.
Summary of a discussion on Burbuja.info - Foro de economía, actualidad y política., translated from Spanish and reviewed before publication. Read the full discussion (31 replies).
More summaries
-
Degussa Review: Buy Gold Tax-Free and Verify Coins
Gold is VAT-free in Spain, unlike silver. Degussa is generally pricey but offers the best deal for Mexican 50 Peso coins. Verifying coins is easier than bullion bars.
-
Putting Spain's El Gordo winnings into JEPQ: €3,920 a month in dividends
Spain's Christmas lottery top prize nets €328,000 after tax. Investing it all in JEPQ promises €3,920 a month, but five-year growth is just 1.39%. Is it worth it?
-
Wheat up 50% from lows as diesel threatens the harvest
Wheat has climbed nearly 50% from its floor, and rising diesel prices threaten the next harvest, with talk of a return to 2022 highs.
-
Bank lifetime annuity at 3%: 33 years to recoup €40,000
€40,000 at 3% takes 33 years to pay back, and fees leave the lifetime annuity at a net 2%, less than a Spanish Treasury bill.
-
Anti-bump lock cylinders: "the lock is only a deterrent"
A debate on KABA, TESA TK100 and Winkhaus blueChip cylinders: some argue that if burglars want to get in, they will come through the wall or another weak point.
-
Cheap prefab homes: the hidden costs behind the low price
Prefabricated houses advertised from 5,000 euros on AliExpress hide costs that triple the final price. Bureaucracy, land, permits and foundations turn the bargain into a headache.