[[Start here]] → [[manage the monkey mind|Mindset]] → Information and Confidence --- ![[more information increases forecast confidence but not accuracy 1.png]] I regretfully remember my highest ever conviction stock pick. It was a little known share that had all the traits of a stock market winner. A highly profitable, inexpensive small cap with a great growth trajectory - a classic GARP share. It was operating in a technological area that seemed inevitable to secure broad adoption - biometric identification. This was way before Apple introduced touch id and then face id. The opportunity and use cases were clear for all to see. My initial purchases were small, but as I researched more, I bought more. I found a community of online investors who agreed. We shared information. The excitement grew as the valuation grew. We even met the CEO, who invited a group of us to lunch at the Mandarin Oriental hotel in London. Perhaps that was the telling moment, but in my naivety I bought even more. The share quadrupled, and grew to 50% of my portfolio. It ended up a fraud. The CEO went to jail. The stock delisted almost worthless. While a painful recollection, it’s an anecdotal observation of what is now a well researched truth. ==The gathering of more information tends to increase the confidence of forecasts, but not their accuracy.== Back in 1973, Paul Slovic released a famous study that quantified this result.[^1] Slovic knew that horse-race handicapping was an uncertain, information game, with strong parallels to investment. So he gathered a group of eight experts in horse-race handicapping, and asked each to choose the 5, 10, 20 and 40 most valuable pieces of information from a set of 88 that could help them predict horse races. They were then asked to judge 40 horse races using these sets of information, first judging with five pieces of information about each horse, then ten and so on. The outcome of this experiment is shown below. ![[slovic-confidence-information-chart.png.png]] Startlingly, prediction confidence grew with information, but accuracy did not. Having more than five pieces of information did nothing to improve predictions. These results have been replicated in many studies and other domains since. Before you stop in your tracks and give up on any extensive stock research, it’s worth reading Philip Tetlock’s “Superforecasters”[^2]. Tetlock found that the top 2% of forecasters are extremely accurate. The difference is that they show some remarkable traits, seeking contrarian views, being independent, recalibrating on new information, unpacking their assumptions and so on. Of course if you think your odds of being in that top 2% are low, there is good news. In investing, you may only need a few pieces of information to make a good decision after all. Many of the greatest [deep value](https://appwww.stockopedia.com/content/deep-value-the-stock-market-strategy-for-the-brave-968629/) investors - including Benjamin Graham, Walter Schloss and others[^3] - were often happy to use just a single piece of information to value each stock (the price of a share relative to its liquidation value). They just backed it up with broad diversification to avoid catastrophic loss. Whether you use one, five or ten pieces of information to select your investments, make sure they are robust and [[expose to return drivers|expose you to key return drivers]]. The important thing is to be consistent, not get cocky and realise that more and more research may not make your favourite stock pick any better. I’ve learned my lesson. It was painful. I’m keen you don’t repeat it! [^1]: [[Slovic - Behavioural problems of adhering to a decision policy]] [^2]: [[Tetlock - Superforecasting]] [^3]: [Deep Value - the stock market strategy for the brave](https://www.stockopedia.com/content/deep-value-the-stock-market-strategy-for-the-brave-968629/)