Risk and honesty

How we make claims

We try hard to publish numbers that are true, and to put the unflattering ones next to the good ones. Concretely:

• Every strategy number comes from a survivorship-free test. Companies that went bankrupt, got acquired or delisted stay in the data at the losses they actually caused. Dropping them is the most common way a backtest flatters itself.
• We test ideas by trying to break them. An idea ships only if it survives being split in half, having its best trades removed, having its settings nudged, and being run on data it was never tuned on. Most of what we try fails, and we do not ship it.
• When a number we already published turns out to be overstated, we correct it downward and say so. That has happened, and it will happen again.
• Where an idea shows no evidence of an edge, we say so, and we say it here rather than burying it. Ignition is the current example, and it is written up below.

What the numbers cannot do

We can measure the past carefully. Nobody can forecast the future, us included. A good forecaster with good instruments still gets rained on, and markets behave worse than weather. Specifically:

• A backtest is not a track record. It is a simulation over data that already happened, not money anyone made.
• One market history. One decade, one sequence of events, which the next is under no obligation to resemble.
• Every strategy here has losing stretches, some long and deep. The drawdowns we publish are real and measured daily.
• Data can be wrong or late. Prices and company details come from third parties.

Ignition is watch-only

Ignition draws a shape. It is not a buy signal, and we have never been able to show that buying it works. The evidence, all survivorship-free:

• Across 77,769 signals, the average 60-day return after a red dot was 1.88%. The same universe over the same windows returned 1.88%. The excess was zero.
• The score that orders the cards does not order outcomes. Its correlation with what happens next is negative, significantly so at 20 days, and the highest-scoring tenth of signals did worse than the lowest-scoring tenth.
• Buying the breakout, best of eleven settings, returned −39% from Dec 2019 to Jun 2026, against +162% for the S&P over the same window.
• The tracked rule is fragile to its own settings. In the version we audited in depth, taking the second-best red dot instead of the best cut the result from 9.8x to 0.4x, moving the holding period five trading days either way cost about two thirds of it, and three trades produced 58% of the total. Across 21 such perturbations it beat the S&P in 7.
• The variant tracked on the page holds one name at a time. Its headline is 14.6x, but that rests on 19 closed trades in a decade, at a worst drawdown of −58%. One company's fate is the whole result.

We publish that rule and its trades as a record kept in the open, not as a recommendation. If it ever turns the corner on live data, the record will show it first and we will say so plainly.

Trading on it

• Only risk money you can afford to lose.
• Size for the drawdown, not the return. Several strategies here hold only a couple of names at a time, and individual companies can go to zero.
• Nothing here needs you to act within minutes.
• If your situation is complicated, a licensed adviser who knows it can help. We do not know it and cannot.

What we are, and are not

InvestCG is a research tool: it publishes what a set of stated rules currently says, and the record of those rules.

• We are not a broker, dealer, investment adviser or financial planner, and nothing here is a personal recommendation or investment, legal or tax advice.
• We never place a trade, hold your money, or touch your account. Every order is one you decide on and enter yourself, at your own broker.
• No relationship here is a fiduciary one.

Responsibility for losses

You are responsible for your own trading decisions and their outcomes. InvestCG, its operators and its contributors are not liable for any loss, direct or indirect, arising from your use of this site, including losses from acting on its signals, numbers or research, from errors or omissions in them, or from the site being wrong, delayed or unavailable. Everything here is provided as is, without warranty of any kind. If you are not willing to accept that, please do not trade on it.

Trading stocks and cryptocurrency involves substantial risk, including the permanent loss of your entire investment.

Questions and corrections

If you think a number here is wrong, tell us. Corrections improve the product and we would rather hear it from you than leave it standing. The FAQ covers how each strategy works and what its record is.