risk rating

5 min readmetropolx

risk
creators

every launch alert comes with a rating of the creator, not of the token. a brand new token has no history, the wallet that launched it often does.

the rating is a score from 0 (safest) to 100 (riskiest), computed the moment the token launches, from every token the same wallet launched before it on nad.fun.

how the rating works

three factors, each turned into a goodness $g$ between 0 (bad) and 1 (good):

factor	question	weight
graduation	do their tokens graduate?	50%
survival	are their tokens still trading a day later?	30%
serial launching	do they spray launches and hope?	20%

the score is what's left of the weighted mean:

\text{score} = 100 \times \left(1 - \frac{\sum_i w_i \, g_i}{\sum_i w_i}\right)

rounded to the nearest integer. a factor without data is left out, and the weights of the others are rescaled.

two rules apply to every factor:

only known outcomes count. a launch counts toward the graduation rate once it graduated or is 24 h old, toward the survival rate once it is 48 h old. a token launched 10 minutes ago tells nothing yet
recent launches weigh more. a launch $t$ days old weighs $w(t) = 0.5^{\,t/60}$ : half as much after 60 days, a quarter after 120

graduation

graduation is when a token leaves the launchpad bonding curve. most tokens never do, the launchpad base rate is about 0.5%.

a creator with 1 graduation out of 2 launches is not a 50% creator, that's mostly luck. so the rate is pulled toward the base rate, as if every creator started with 20 average launches behind them:

p_\text{grad} = \frac{G + 0.005 \times 20}{N + 20}

with $G$ the weighted count of graduated launches and $N$ the weighted count of resolved ones. the more launches a creator has, the more their own record wins over the base rate.

rates span from 0.03% to 50%, so goodness is read on a log scale, from 0 at 0.1% to 1 at 15%:

g_\text{grad} = \operatorname{clamp}_{[0,1]}\left(\frac{\ln(p_\text{grad} / 0.001)}{\ln(0.15 / 0.001)}\right)

a graduation after a community takeover (CTO) does not count: the community did the work, not the launcher.

survival

a launch survived if it still traded a day later: it has a daily candle opening at least 24 h after launch. the base rate is about 21%, smoothed the same way with 10 pseudo-launches, then read on a linear scale from 0 at 5% to 1 at 60%:

p_\text{surv} = \frac{S + 0.21 \times 10}{N_s + 10} \qquad g_\text{surv} = \operatorname{clamp}_{[0,1]}\left(\frac{p_\text{surv} - 0.05}{0.60 - 0.05}\right)

launches without candle data are skipped. when none have any, survival is left out of the score.

serial launching

some wallets launch dozens of tokens a week and walk away from each. this factor counts the launches of the last 30 days that did not graduate, $F$ , discounted by the graduation record: a creator who does graduate tokens is forgiven their misses.

F' = F \times (1 - g_\text{grad}) \qquad g_\text{serial} = 1 - \operatorname{clamp}_{[0,1]}\left(\frac{\ln(1 + F')}{\ln(1 + 30)}\right)

log scale again: the first few misses cost the most, and 30 adjusted misses take this factor to 0.

rating levels

tier	score	alert border	discord feed
low	0 – 35	🟦 blue `#54A9FF`	🟦 bluechip
medium	36 – 64	🟨 yellow `#FFE254`	🟧 medium
high	65 – 100	🟥 red `#FF5454`	🟥 high
unknown	–	⬜ grey `#95A5A6`	⬛ clusterfuck only

every rated launch also goes to ⬛ clusterfuck, see add to discord.

graduations carry no rating, they get their own color and feed:

event	alert border	discord feed
graduation	🟪 purple `#6E54FF`	🟪 graduates

colors

the graduation purple is the metropolx primary, $\texttt{\#6E54FF} = \text{hsl}(249.1°,\ 100\%,\ 66.5\%)$ . the tier colors keep its saturation and lightness and only turn the hue:

color	hue	hsl	hex
🟪 purple (primary)	249.1°	hsl(249.1°, 100%, 66.5%)	`#6E54FF`
🟦 blue	210°	hsl(210°, 100%, 66.5%)	`#54A9FF`
🟨 yellow	50°	hsl(50°, 100%, 66.5%)	`#FFE254`
🟥 red	0°	hsl(0°, 100%, 66.5%)	`#FF5454`

examples

four creators, their earlier launches all a few days old:

record	graduation	survival	serial	score	tier
2 graduated, 4 survived, out of 5	0.88	0.64	0.91	19	low
1 graduated, 3 survived, out of 10	0.72	0.37	0.63	41	medium
0 graduated, 0 survived, out of 10	0.24	0.10	0.37	77	high
0 graduated, 1 survived, out of 40	0.11	0.02	0.00	94	high

note the second row: 1 graduation in 10 is 20× the launchpad base rate, that alone keeps the creator out of the high tier.

first-time creators

no track record is a red flag, not a neutral one. a creator with no resolved launch is always rated high, with a score of at least 85:

a wallet launching its very first token
a wallet whose earlier launches are all younger than 24 h and none graduated, e.g. three launches in the last hour

fresh wallets cost nothing to make, so a rug often comes from one.

unknown

unknown means we could not read the creator's history in time (database error, or the 2 s lookup timeout). freshness beats completeness: the alert goes out anyway, in grey, to clusterfuck only.

limitations

it rates a wallet, not a person. a serial rugger with a fresh wallet looks like any first-time creator, which is why those are rated high
the past is not the future. a creator with a clean record can rug next time
nad.fun only. launches on other launchpads or chains are not counted
last 500 launches. older history of very prolific wallets is ignored, it weighs almost nothing anyway after the 60-day half-life
candle gaps. survival depends on daily candles; tokens we have no candles for are skipped, not counted as dead

not financial advice. the rating is one signal, read it with the creator metrics shown next to it.