Inputs:
Series ( NumericSeries ),
Len ( NumericSimple );

Variables:
TEMA0 ( 0 ),
TEMA1 ( 0 );

TEMA0 = TEMA(Series, Len);
TEMA1 = TEMA(TEMA0, Len);

ZeroLagTEMA = TEMA0 + (TEMA0 - TEMA1);

Inputs:
Series ( NumericSeries ),
Len ( NumericSimple );

Variables:
TEMA0 ( 0 ),
TEMA1 ( 0 ),
ZL (0);

TEMA0 = TEMA(Series, Len);
TEMA1 = TEMA(TEMA0, Len);

ZL = TEMA0 + (TEMA0 - TEMA1);
ZeroLagTEMA = TEMA(ZL, Len);

Inputs:
Series ( NumericSeries ),
Len ( NumericSimple );

Variables:
DEMA0 ( 0 ),
DEMA1 ( 0 );

DEMA0 = DEMA(Series, Len);
DEMA1 = DEMA(DEMA0, Len);

ZeroLagDEMA = DEMA0 + (DEMA0 - DEMA1);

Inputs:
Series ( NumericSeries ),
Len ( NumericSimple );

Variables:
DEMA0 ( 0 ),
DEMA1 ( 0 ),
ZL ( 0 );

DEMA0 = DEMA(Series, Len);
DEMA1 = DEMA(DEMA0, Len);

ZL = DEMA0 + (DEMA0 - DEMA1);
ZeroLagDEMA = DEMA(ZL, Len);

A short answer first:

There is no such thing as a ZeroLag moving average. The only indicator that has zerolag is price itself, unless you make an assumption on the underlying data series to be a perfect trendline or sinoidal cycle.

The moving average produces smoothing and pays for it by introducing a lag.

The various methods to reduce the lag produce low lag moving averages as they should be called correctly. Possibly there is an infinite variety of low lag moving averages, so there cannot be any correct formula.

You can test on your chart, whether the above moving averages are low lag moving averages or not.

I bet you tell children there is no Santa Clause either...

There really used to be an Easter Bunny... but Santa Clause ate it for Christmas Dinner...