How to use the Titien-Moore criterion?

Question

Python

There are a number of digital values ​​of length n. There may be several anomalous values ​​in this series (according to the maximum value). Tell me how to use the Titien-Moore test to determine these anomalous values. You can use the Smirnov-Grubbs criterion, it is very simple:

import math

def rem_anomalies(l):
  alpha = 2.447
  avg = sum(l) / len(l)
  sig = 0
  for i in l:
    sig += (i - avg) * (i - avg)
  sig = math.sqrt(sig/len(l))

  l2 = []
  for i in l:
    if alpha > (i - avg)/sig:
      l2.append(i)

  return l2

l = [1, 22, 123, 121, 11, 23, 1223, 11,43 ,12, 123, 142, 11, 2333, 121, 123]
print(l)
print(rem_anomalies(l))

> [1, 22, 123, 121, 11, 23, 1223, 11, 43, 12, 123, 142, 11, 121, 123]
But as we can see, it does not fit for several anomalies (in this example there were two anomalies - 2333, which the algorithm was able to determine, and 1223, which we missed)
I am not strong in algorithms, so if anyone knows and understands how to apply the Titien-Moore criterion or knows other criteria for finding several anomalous values, please tell me as simply as possible. Desirable (but not required) using pseudocode example 8)

Answer 1

The description of the algorithm and the logarithm critical value function was taken from here
JS example: https://jsfiddle.net/cs0m7r5h/11/

Answer 2

There is a median filter to get rid of outliers.
More precisely, anomalies are not removed from the series, but are replaced by average values.